diff --git a/.lfs-assets-id b/.lfs-assets-id index 7d52a573a..9c701c8f1 100644 --- a/.lfs-assets-id +++ b/.lfs-assets-id @@ -1,4 +1,5 @@ 00e254e3cc6879b3c08cfc5a96ad882a6b1b0a232afcff4934bc1555e05f7168 +00f02e01fabb70c689cecb53dcbda63288279456e9fed68e637d190a9bb917db 017e1ff56435818b639cc788ee35df6cee7ac2aea19cc62dcb10a3a0c60b6340 01fa62d89cd626cebaed7a6975a9488848c1764e4a6ef4798597894c93a1379c 030e8ed8bc25b858a391eb1f9eb275c144de9bbe08ce0e5150c068903d8f8246 @@ -41,9 +42,12 @@ 1d64387aa4bea78bd412ea3892bf17f48b32856074dfe3b5105f0bbc1b15603f 1ee8cf9693351db7afe68f6fc32942845caae3e1030c688efa6c5d0b24229f46 1f5cf829d2e58032e5d9067f1e7bf3fe7644cd5fe2825c81ef7fbaa445f496a7 +1f69f25b0168c39018c214ae39f2bb8fe6da97e1df389c2cb88cf9bde2f08ace 21d506c397dfeb3edbbbc253b923f59be6edb516689677ebef535296c6c62242 +22053c4cdaaaf6169d90e4ee03f8b994cb682281ab6787bf14c617301db5663b 22baa768886091d61e13e894610dfea3435dfd201f8300fbebe46cd6cf814c0b 22c0fae7298efcb4566c4d4cc19a53c505476cccf6518328303e2b7334110781 +244c05ca33384bcdd519ff88fad036ae7140a5438b76a8248401cdc065582ac7 24776e0eb2ca28529d8913612c14fdb6eb822bd991628272034b3c3176b42c5b 24829ec04c7bcf2a3681c929b9b0b9315d44f6557c6814c4bc398fc21617faf0 2588044f47a5fd80b2bacc4ae4c19cd32565dd9b53d3e00deae385016e2ed4f9 @@ -57,6 +61,7 @@ 2bdda1022a866fd7fa6307e667173c821c206ae007308fb54cf580d668e01e6c 2c4393bc88f742dc9cd253c7167d882164671df75195a19d9d79af23f1c1e351 2cc1e040bdc20737674a00039cd80e16758c511e697e524ed17e4335ec93e9a6 +2ced489a1f314abe77f99f022eb34ad496e47e83e90a1d10947ea633aee367ac 2d0ec33c9c9e8cdc9560d3b890f5bdb16a96dd22620ae80902393715e114e18d 2d1386a66ab4548b5bdfc95581a77d5fd21852094da5ad2a7844b04e47e1e612 2e1c31a2c560f3539be5bda1084939765c80ddd814bd6919fd8a3465f9b8aee6 @@ -121,6 +126,8 @@ 4f7c486b471f2a5dcd94c97f52089b487c3f4aa1bbee14c3ac19472e4144b2eb 4f8aaeefb1885b1f0ff918e75a0bc63fc2f900fbb44a0bd24e15387a69f008d3 500e238667025845f2d31d0d1d814ba5dca5b34f02aa3a703c3fffbd1c67d91a +5079ad94ed5d03057e754b97692f1675603d515ebd663fcf34b87c45103e449c +508c08fa0f4bc7554436252e43e720240c46e6a466bd43ac46db44167c9fa476 519d15a29a75c6449b8492b38f7f626865da61a392c1b69a1aa51ce2cf5a802a 52570e7b3d0066899283c3ef1fbb4241f64ad3d0a1bd6c9cc179ebe2262c8e86 52ac257854fadc38544098fc62402fee94e38a4688bf1c44fe34ee6cffd4784e @@ -159,6 +166,7 @@ 63528b5c9828d4d26240ac2ed20cdae7ff8e51fb742e622aa55fedae094ef9b4 6374215d9c908c17b39585aec40a23dbd06c403379d86fba4fae6bdb50fbbd50 638db3e132b98e5bdaf919592ab3cc9d5c2fc11a4360dfd2a59cebdd51225551 +63b8bf529647b3ffaa062a0f8ee7d9cb05e7c8c5423d11c1f3e907a39224a39b 6402396156d04f7269074f4e58fa40ad1147f9298179bf388962d6ef9f916824 642076c10a6fdecb53994f6dde89cf74b445e23445035378e914c36b95cdc4f0 64378cabaecdb50e2e1e644c1d6bc359f0627ebe313870ad09e1e41ed972529b @@ -184,6 +192,7 @@ 6f78849094e49f3f4d9b84a7bb76c50e1c86cecd9f558a12f7698b31efbdfd79 6fe281722cd1cf4da63248e09b84f5c11ddf73f13ef29e03e955cae8036548e3 705c335a0d7edb3edf247ea313f2ae334086ce3e6bf4e1b64a8b306befd1c816 +713cf4ebab2cc2c3ebbc31fa42159f16df41a345eb8588e11b680dafc46b3103 7164f21d0c4a5bd5fdfcab5918cd1e070314750bdfab0b9a32c2c7f1b3df5e87 719efc2cb0f94f49721de58ec6c9dc58bdbb16b241d00c61eafe47f737fcbfc2 71e8fb79b34ffbae9dacba02367d940f8d2397ff9a7687c2fc70cd1b16209b08 @@ -249,6 +258,7 @@ 9266fa48c776ff241b665769a932ecf184defed1ccbb47b87a0930c78dedc7ab 937b0a23076f948343b8b7ac005bc70ea7080c391218fb4896462699b8cb59c4 93a4486c3ad84002ae5506dee5827e2da3c03660b5fa28efab2975af81f8e303 +93e819f4b2d26df1cd5eb9805bd2fe15767941c6c8d66c58eba86d9249cb8ce1 94abf897fcd6d93b0f78ea309b10ef5b83a8b95e46aa29974a43dbefb3741c0c 955f006886606699ceab5fb38693e4fe10f74f450a2fd6dfd75af09b8b134d31 958399e198a323bd5653838154185172ed3ef7b097f7450ccf5c3a9be2639887 @@ -269,6 +279,7 @@ 9a01dfeadf8fbc8679b9be7635900dae664bad008309d2fdf18570cb9b17f75d 9a3bd7573b3d0379ccf288c6284298a71f9a351d70b996ede69b5a97e883d94d 9ae5108706e2d98e5fe27f4bcbe1cc73183be680bb94257989b26da6522a729a +9c18fe91ba5401cab754456a9bfbfe54a6c307b0d4b880259fb4218edef77998 9c2b046e90d8d62e1446b9d2287de6c751eb74a8715a9e108e89b5b875ce1100 9def90abb62c1b9872ae2dbed3e425850ae44f98d3664e8b29e2444548e9fec9 9e1ec477e8af8356c3b731f8815f19b57bb404bd7c1629a2020bc9d90b0c028a @@ -276,6 +287,7 @@ a15ad0797f0d445cafcf5afb14f26df0aee2417181a2081ad26b1c10e0aaf79c a2b0fbfc6f24cb86e3c612be4d59f5dc48b4e3e73620b480dd9f54dccf4d90da a39fc5f054db506e83a4a4ec47eba1f7f9bf9bfdc983174e699312f32f42f1f5 +a3a7bf32a44deebb795cf6faf6fc88936d24fc94157468f352c8972906318984 a4e3dfadeb34bba861ef56046d2fc99c4d50d8475882d610a0b652d39c510f6b a551e2941365201552a3a819f035c64f297467464f7c7c349d00711316ce3c57 a55638ba902c8afb52d5b006f2ac438c72dfcdd6325efe7488b125f7a9662989 @@ -300,6 +312,7 @@ b095f9d384fbac256b5870d3ca01d687e776a81457351a8a6fa56e7f1d09c4b0 b0b6b33338e507bbc056f4d746e705b3bc8f5bb4cc4449187d69997797f91a1d b0ba1f16f688d6af7323b4747c6fbfa88aa07cb3e4497d97cee120b9fed85415 b258b43c18558bfcd7f68e77c68125c1ccb6b3f17556273a1abdaab712f92e21 +b3845ddf3afff808cd6449daa546caf4d8b18f109b7ebb2344990e4c37fea291 b3e7072ef46bb1cdf57d4d9189fbffd3980fa968296a6c58f66e0a11f552d8de b41780978c344f29cc340883e41610b4f2364845625d1d1e6f6c4e4510214b60 b51ab4333dc1c5a1893ef09c7cad75b97c0873648cfef1902e5b5d67c5774330 @@ -344,6 +357,7 @@ cd86c9748b1740e51e852419ea4da91d9a1488cdc8d890cffba3d409e3147594 cd9fc8d6cdc07793f16b8455caef2522528727a40e0ecc1d4f83cf506769b1be ce4982cba30bd5aaa352e51d55e1cf56c56ae6d573d644f36257de1177717228 ced18f246d507c049241f107d4e07e77a1d0afa4ae2154e099b9a3c286cb7141 +cf266818a2ab4d185db9c21d28019871604dee7b8755ff27b7c6ebc41c908004 cf43c33f24c1defbdfa241ffcc4e573bf548208ebb1bf05b54256d191c5dba46 d00d1522a06b5b67a0aea355c268ffe575e4a28651fa4fbf0e204705be3ce47b d00fd6bac2eda0821a04ff10277663d211127745a19a474c0feee41a16914fdc @@ -388,6 +402,7 @@ ebd2c0505b95df7c3fbcd132fe97d35cf85dc6be9b939f26cb0962f8134a5198 ec7968db175dda2a12e69f7d5688658210ed9c8bf484391c653d28b4b3c068e7 eca2922ecc84ba0643f1ca828b866503d16f128e9ca8155d645bdab293f6d69f ecbabaa5a989a8c11f5a57e874c135f386da5379d6689ae6d5d4264a652fa5e0 +ed91959038654fe3a02da9d51ab0d8f5187003b899c767a2de436f6b0808425e edf46fde5a137ffc264b86333ff655d7699de3d69a8bd401fc72d50a78d238c8 efe8cec1c28bdc92461bb828d2d0a1336c6d34526c6e9133ab23baa644ebaa0a f044290716bfa54d669c70b73a5380833bc00fb0a4764659847e2fa52481199d @@ -404,11 +419,14 @@ f6a4579c6f7dcee7a0277e9e53b7df965191a3f80a30521a750dc9223d93f561 f7d87339bd5bdb369062a4f658e4d9f6ee015ad7fa7b84c45a00412f10cb2199 f872451188e95b8ed0c32e2f250c8ed80205e66f828b8aca8bc34ee17b0a929a f8ffe4f9920dc63fb85b864cc1afffb66e45c6e302d175905fd7db04ba67b8c6 +f901a05ccbd6bd1701b83ea8baa1c454be72cb8c41b00e2cec1bb09d6154e691 faa818f5b917024c1740d80d450c92bb982b4f16aa02e8727250c96422a4093e fadc9184004a8e2daf3abec9426a9224241cdece35be388870fd17b3a9ba54a4 fb73571feabc1f0c10dc749acb28bf2d0efceeedca2115eafd79ad5b82cdcba0 fb8bc76eafa39558b6044907f23416faf99099cefb8afd06a2cf5afbdae1b506 +fbee2fea90f1e1ce3251869a84b87723e37ab6a512c515a4fb6b9f9d72e5ee0c fc2232345c7145c1181090555d7a76918dde6464239a04cc7b36b98674212245 +fc3565edb4c6e03a7758fbe3c25e5805098a6642811c6f67f060b093f9fcc6a4 fc59b38546bd80f92ac7e6b06f418808ee62097d5452c62aff607f792ffd529a fc5d76439885e717a9f0c4e445cdb9de9b566416200b131e63ba99a9f0fcb3cd fde2fbc5d0ebf03b1375d4a2d05ae4ae877e2ef72989b9aecde9b963fa004025 diff --git a/datasets/_deprecated_195_auto_price/README.md b/datasets/_deprecated_195_auto_price/README.md index d3209704e..f69720fbe 100644 --- a/datasets/_deprecated_195_auto_price/README.md +++ b/datasets/_deprecated_195_auto_price/README.md @@ -1,6 +1,6 @@ -# 195_auto_price +# _deprecated_195_auto_price -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/195_auto_price.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_195_auto_price.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_195_auto_price/195_auto_price.tsv.gz b/datasets/_deprecated_195_auto_price/_deprecated_195_auto_price.tsv.gz similarity index 100% rename from datasets/_deprecated_195_auto_price/195_auto_price.tsv.gz rename to datasets/_deprecated_195_auto_price/_deprecated_195_auto_price.tsv.gz diff --git a/datasets/_deprecated_195_auto_price/metadata.yaml b/datasets/_deprecated_195_auto_price/metadata.yaml index 260759b25..dfc8dd57a 100644 --- a/datasets/_deprecated_195_auto_price/metadata.yaml +++ b/datasets/_deprecated_195_auto_price/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: 195_auto_price +dataset: _deprecated_195_auto_price description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -87,4 +87,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_195_auto_price/summary_stats.tsv b/datasets/_deprecated_195_auto_price/summary_stats.tsv index e370ecbd0..a9f2838d5 100644 --- a/datasets/_deprecated_195_auto_price/summary_stats.tsv +++ b/datasets/_deprecated_195_auto_price/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -195_auto_price 159 15 0 1 14 continuous 145.0 0.0005037819354895438 regression +_deprecated_195_auto_price 159 15 0 1 14 continuous 145.0 0.0005037819354895438 regression diff --git a/datasets/_deprecated_207_autoPrice/README.md b/datasets/_deprecated_207_autoPrice/README.md index ea491cd97..013150c56 100644 --- a/datasets/_deprecated_207_autoPrice/README.md +++ b/datasets/_deprecated_207_autoPrice/README.md @@ -1,6 +1,6 @@ -# 207_autoPrice +# _deprecated_207_autoPrice -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/207_autoPrice.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_207_autoPrice.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_207_autoPrice/207_autoPrice.tsv.gz b/datasets/_deprecated_207_autoPrice/_deprecated_207_autoPrice.tsv.gz similarity index 100% rename from datasets/_deprecated_207_autoPrice/207_autoPrice.tsv.gz rename to datasets/_deprecated_207_autoPrice/_deprecated_207_autoPrice.tsv.gz diff --git a/datasets/_deprecated_207_autoPrice/metadata.yaml b/datasets/_deprecated_207_autoPrice/metadata.yaml index d46df57e2..d215ab268 100644 --- a/datasets/_deprecated_207_autoPrice/metadata.yaml +++ b/datasets/_deprecated_207_autoPrice/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: 207_autoPrice +dataset: _deprecated_207_autoPrice description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -87,4 +87,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_207_autoPrice/summary_stats.tsv b/datasets/_deprecated_207_autoPrice/summary_stats.tsv index c680bc12b..8e45393f5 100644 --- a/datasets/_deprecated_207_autoPrice/summary_stats.tsv +++ b/datasets/_deprecated_207_autoPrice/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -207_autoPrice 159 15 0 1 14 continuous 145.0 0.0005037819354895438 regression +_deprecated_207_autoPrice 159 15 0 1 14 continuous 145.0 0.0005037819354895438 regression diff --git a/datasets/_deprecated_australian/README.md b/datasets/_deprecated_australian/README.md index cb93bac2e..100d0db19 100644 --- a/datasets/_deprecated_australian/README.md +++ b/datasets/_deprecated_australian/README.md @@ -1,6 +1,6 @@ -# australian +# _deprecated_australian -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/australian.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_australian.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_australian/australian.tsv.gz b/datasets/_deprecated_australian/_deprecated_australian.tsv.gz similarity index 100% rename from datasets/_deprecated_australian/australian.tsv.gz rename to datasets/_deprecated_australian/_deprecated_australian.tsv.gz diff --git a/datasets/_deprecated_australian/metadata.yaml b/datasets/_deprecated_australian/metadata.yaml index 5b6d90847..9897a48f4 100644 --- a/datasets/_deprecated_australian/metadata.yaml +++ b/datasets/_deprecated_australian/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: australian +dataset: _deprecated_australian description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -82,4 +82,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_australian/summary_stats.tsv b/datasets/_deprecated_australian/summary_stats.tsv index 1d380761e..cd300e583 100644 --- a/datasets/_deprecated_australian/summary_stats.tsv +++ b/datasets/_deprecated_australian/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -australian 690 14 4 3 7 categorical 2.0 0.012131905061961762 classification +_deprecated_australian 690 14 4 3 7 categorical 2.0 0.012131905061961762 classification diff --git a/datasets/_deprecated_auto/README.md b/datasets/_deprecated_auto/README.md index f96ca0e70..a8e07af7b 100644 --- a/datasets/_deprecated_auto/README.md +++ b/datasets/_deprecated_auto/README.md @@ -1,6 +1,6 @@ -# auto +# _deprecated_auto -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/auto.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_auto.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_auto/auto.tsv.gz b/datasets/_deprecated_auto/_deprecated_auto.tsv.gz similarity index 100% rename from datasets/_deprecated_auto/auto.tsv.gz rename to datasets/_deprecated_auto/_deprecated_auto.tsv.gz diff --git a/datasets/_deprecated_auto/metadata.yaml b/datasets/_deprecated_auto/metadata.yaml index e341f80ff..683b4f3eb 100644 --- a/datasets/_deprecated_auto/metadata.yaml +++ b/datasets/_deprecated_auto/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: auto +dataset: _deprecated_auto description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -137,4 +137,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_auto/summary_stats.tsv b/datasets/_deprecated_auto/summary_stats.tsv index 552fb31d5..986642898 100644 --- a/datasets/_deprecated_auto/summary_stats.tsv +++ b/datasets/_deprecated_auto/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -auto 202 25 3 6 16 categorical 5.0 0.045375453386922844 classification +_deprecated_auto 202 25 3 6 16 categorical 5.0 0.045375453386922844 classification diff --git a/datasets/_deprecated_breast/README.md b/datasets/_deprecated_breast/README.md index f77d676e7..412d4e1df 100644 --- a/datasets/_deprecated_breast/README.md +++ b/datasets/_deprecated_breast/README.md @@ -1,6 +1,6 @@ -# breast +# _deprecated_breast -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/breast.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_breast.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_breast/breast.tsv.gz b/datasets/_deprecated_breast/_deprecated_breast.tsv.gz similarity index 100% rename from datasets/_deprecated_breast/breast.tsv.gz rename to datasets/_deprecated_breast/_deprecated_breast.tsv.gz diff --git a/datasets/_deprecated_breast/metadata.yaml b/datasets/_deprecated_breast/metadata.yaml index f70b92a95..2f12c12ab 100644 --- a/datasets/_deprecated_breast/metadata.yaml +++ b/datasets/_deprecated_breast/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: breast +dataset: _deprecated_breast description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -62,4 +62,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_breast/summary_stats.tsv b/datasets/_deprecated_breast/summary_stats.tsv index f16e3454f..18a0aaf2d 100644 --- a/datasets/_deprecated_breast/summary_stats.tsv +++ b/datasets/_deprecated_breast/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -breast 699 10 0 8 2 categorical 2.0 0.09637516091862275 classification +_deprecated_breast 699 10 0 8 2 categorical 2.0 0.09637516091862275 classification diff --git a/datasets/_deprecated_breast_cancer_wisconsin/README.md b/datasets/_deprecated_breast_cancer_wisconsin/README.md index 86772b591..93c48ddd4 100644 --- a/datasets/_deprecated_breast_cancer_wisconsin/README.md +++ b/datasets/_deprecated_breast_cancer_wisconsin/README.md @@ -1,6 +1,6 @@ -# breast_cancer_wisconsin +# _deprecated_breast_cancer_wisconsin -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/breast_cancer_wisconsin.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_breast_cancer_wisconsin.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_breast_cancer_wisconsin/breast_cancer_wisconsin.tsv.gz b/datasets/_deprecated_breast_cancer_wisconsin/_deprecated_breast_cancer_wisconsin.tsv.gz similarity index 100% rename from datasets/_deprecated_breast_cancer_wisconsin/breast_cancer_wisconsin.tsv.gz rename to datasets/_deprecated_breast_cancer_wisconsin/_deprecated_breast_cancer_wisconsin.tsv.gz diff --git a/datasets/_deprecated_breast_cancer_wisconsin/metadata.yaml b/datasets/_deprecated_breast_cancer_wisconsin/metadata.yaml index 548bce50c..20a7054ac 100644 --- a/datasets/_deprecated_breast_cancer_wisconsin/metadata.yaml +++ b/datasets/_deprecated_breast_cancer_wisconsin/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: breast_cancer_wisconsin +dataset: _deprecated_breast_cancer_wisconsin description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -162,4 +162,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_breast_cancer_wisconsin/summary_stats.tsv b/datasets/_deprecated_breast_cancer_wisconsin/summary_stats.tsv index 8663df857..dda361a03 100644 --- a/datasets/_deprecated_breast_cancer_wisconsin/summary_stats.tsv +++ b/datasets/_deprecated_breast_cancer_wisconsin/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -breast_cancer_wisconsin 569 30 0 0 30 categorical 2.0 0.06493987849061501 classification +_deprecated_breast_cancer_wisconsin 569 30 0 0 30 categorical 2.0 0.06493987849061501 classification diff --git a/datasets/_deprecated_breast_w/README.md b/datasets/_deprecated_breast_w/README.md index 334b5f0de..594bac19d 100644 --- a/datasets/_deprecated_breast_w/README.md +++ b/datasets/_deprecated_breast_w/README.md @@ -1,6 +1,6 @@ -# breast_w +# _deprecated_breast_w -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/breast_w.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_breast_w.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_breast_w/breast_w.tsv.gz b/datasets/_deprecated_breast_w/_deprecated_breast_w.tsv.gz similarity index 100% rename from datasets/_deprecated_breast_w/breast_w.tsv.gz rename to datasets/_deprecated_breast_w/_deprecated_breast_w.tsv.gz diff --git a/datasets/_deprecated_breast_w/metadata.yaml b/datasets/_deprecated_breast_w/metadata.yaml index 7205f2290..8bc062914 100644 --- a/datasets/_deprecated_breast_w/metadata.yaml +++ b/datasets/_deprecated_breast_w/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: breast_w +dataset: _deprecated_breast_w description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -57,4 +57,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_breast_w/summary_stats.tsv b/datasets/_deprecated_breast_w/summary_stats.tsv index 1d9e5970b..3c48de60a 100644 --- a/datasets/_deprecated_breast_w/summary_stats.tsv +++ b/datasets/_deprecated_breast_w/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -breast_w 699 9 0 8 1 categorical 2.0 0.09637516091862275 classification +_deprecated_breast_w 699 9 0 8 1 categorical 2.0 0.09637516091862275 classification diff --git a/datasets/_deprecated_buggyCrx/README.md b/datasets/_deprecated_buggyCrx/README.md index 9f6b2f820..6524e6bf7 100644 --- a/datasets/_deprecated_buggyCrx/README.md +++ b/datasets/_deprecated_buggyCrx/README.md @@ -1,6 +1,6 @@ -# buggyCrx +# _deprecated_buggyCrx -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/buggyCrx.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_buggyCrx.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_buggyCrx/buggyCrx.tsv.gz b/datasets/_deprecated_buggyCrx/_deprecated_buggyCrx.tsv.gz similarity index 100% rename from datasets/_deprecated_buggyCrx/buggyCrx.tsv.gz rename to datasets/_deprecated_buggyCrx/_deprecated_buggyCrx.tsv.gz diff --git a/datasets/_deprecated_buggyCrx/metadata.yaml b/datasets/_deprecated_buggyCrx/metadata.yaml index d4940ee5c..ef70bd1d9 100644 --- a/datasets/_deprecated_buggyCrx/metadata.yaml +++ b/datasets/_deprecated_buggyCrx/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: buggyCrx +dataset: _deprecated_buggyCrx description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -87,4 +87,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_buggyCrx/summary_stats.tsv b/datasets/_deprecated_buggyCrx/summary_stats.tsv index c3483c352..64b42f221 100644 --- a/datasets/_deprecated_buggyCrx/summary_stats.tsv +++ b/datasets/_deprecated_buggyCrx/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -buggyCrx 690 15 3 5 7 categorical 2.0 0.012131905061961762 classification +_deprecated_buggyCrx 690 15 3 5 7 categorical 2.0 0.012131905061961762 classification diff --git a/datasets/_deprecated_car/README.md b/datasets/_deprecated_car/README.md index ad010ba93..1584c2693 100644 --- a/datasets/_deprecated_car/README.md +++ b/datasets/_deprecated_car/README.md @@ -1,6 +1,6 @@ -# car +# _deprecated_car -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/car.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_car.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_car/car.tsv.gz b/datasets/_deprecated_car/_deprecated_car.tsv.gz similarity index 100% rename from datasets/_deprecated_car/car.tsv.gz rename to datasets/_deprecated_car/_deprecated_car.tsv.gz diff --git a/datasets/_deprecated_car/metadata.yaml b/datasets/_deprecated_car/metadata.yaml index e78700405..5d5519398 100644 --- a/datasets/_deprecated_car/metadata.yaml +++ b/datasets/_deprecated_car/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by Praneel Chakraborty -dataset: car +dataset: _deprecated_car description: > dataset for car evaluation based on six factors; factors considered are buying price, maintenance price, # doors, # people carriable, size of boot, and estimated safety @@ -11,36 +11,40 @@ publication: > In 8th Intl Workshop on Expert Systems and their Applications, Avignon, France. 1988. task: classification keywords: - - mobile vehicle - - expense + - + - target: type: categorical - description: accceptability of the car - code: 0 = acceptable, 1 = good, 2 = unacceptable, 3 = very good -features: # list of features in the dataset + description: None yet. See our contributing guide to help us add one. + code: None yet. See our contributing guide to help us add one. +features: - name: buying type: categorical - description: buying price of car (vhigh, high, med, low) - code: 3 = vhigh, 0 = high, 2 = med, 1 = low + description: # optional but recommended, what the feature measures/indicates, unit + code: # optional, coding information, e.g., Control = 0, Case = 1 + transform: # optional, any transformation performed on the feature, e.g., log scaled - name: maint type: categorical - description: price of maintenance - code: 3 = vhigh, 0 = high, 2 = med, 1 = low + description: + code: + transform: - name: doors type: categorical - description: number of doors - code: 0 = 2 doors, 1 = 3 doors, 2 = 4 doors, 3 = 5+ doors + description: + code: + transform: - name: persons type: categorical - description: capacity (passengers + driver) - code: 0 = 2 people, 1 = 4 people, 2 = 5+ people + description: + code: + transform: - name: lug_boot type: categorical - description: size of luggage boot/trunk - code: 2 = small, 1 = medium, 0 = big + description: + code: + transform: - name: safety type: categorical - description: estimated safety rating - code: 1 = low, 2 = med, 0 = high - deprecated: true - \ No newline at end of file + description: + code: + transform: diff --git a/datasets/_deprecated_car/summary_stats.tsv b/datasets/_deprecated_car/summary_stats.tsv index 0fca1740b..b75639dac 100644 --- a/datasets/_deprecated_car/summary_stats.tsv +++ b/datasets/_deprecated_car/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -car 1728 6 0 6 0 categorical 4.0 0.39028831590077734 classification +_deprecated_car 1728 6 0 6 0 categorical 4.0 0.39028831590077734 classification diff --git a/datasets/_deprecated_car_evaluation/README.md b/datasets/_deprecated_car_evaluation/README.md index e2eafe3a9..deeba0a8c 100644 --- a/datasets/_deprecated_car_evaluation/README.md +++ b/datasets/_deprecated_car_evaluation/README.md @@ -1,6 +1,6 @@ -# car_evaluation +# _deprecated_car_evaluation -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/car_evaluation.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_car_evaluation.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_car_evaluation/car_evaluation.tsv.gz b/datasets/_deprecated_car_evaluation/_deprecated_car_evaluation.tsv.gz similarity index 100% rename from datasets/_deprecated_car_evaluation/car_evaluation.tsv.gz rename to datasets/_deprecated_car_evaluation/_deprecated_car_evaluation.tsv.gz diff --git a/datasets/_deprecated_car_evaluation/metadata.yaml b/datasets/_deprecated_car_evaluation/metadata.yaml index 14e72df96..680767eba 100644 --- a/datasets/_deprecated_car_evaluation/metadata.yaml +++ b/datasets/_deprecated_car_evaluation/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: car_evaluation +dataset: _deprecated_car_evaluation description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -117,4 +117,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_car_evaluation/summary_stats.tsv b/datasets/_deprecated_car_evaluation/summary_stats.tsv index 098e0428c..879dc2e24 100644 --- a/datasets/_deprecated_car_evaluation/summary_stats.tsv +++ b/datasets/_deprecated_car_evaluation/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -car_evaluation 1728 21 21 0 0 categorical 4.0 0.39028831590077734 classification +_deprecated_car_evaluation 1728 21 21 0 0 categorical 4.0 0.39028831590077734 classification diff --git a/datasets/_deprecated_cleve/README.md b/datasets/_deprecated_cleve/README.md index 11ca03e0c..090e5cb6c 100644 --- a/datasets/_deprecated_cleve/README.md +++ b/datasets/_deprecated_cleve/README.md @@ -1,6 +1,6 @@ -# cleve +# _deprecated_cleve -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/cleve.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_cleve.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_cleve/cleve.tsv.gz b/datasets/_deprecated_cleve/_deprecated_cleve.tsv.gz similarity index 100% rename from datasets/_deprecated_cleve/cleve.tsv.gz rename to datasets/_deprecated_cleve/_deprecated_cleve.tsv.gz diff --git a/datasets/_deprecated_cleve/metadata.yaml b/datasets/_deprecated_cleve/metadata.yaml index 2512c4ef9..8130a4ece 100644 --- a/datasets/_deprecated_cleve/metadata.yaml +++ b/datasets/_deprecated_cleve/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: cleve +dataset: _deprecated_cleve description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -77,4 +77,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_cleve/summary_stats.tsv b/datasets/_deprecated_cleve/summary_stats.tsv index cee950e0b..392e9744f 100644 --- a/datasets/_deprecated_cleve/summary_stats.tsv +++ b/datasets/_deprecated_cleve/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -cleve 303 13 3 5 5 categorical 2.0 0.007940398000196063 classification +_deprecated_cleve 303 13 3 5 5 categorical 2.0 0.007940398000196063 classification diff --git a/datasets/_deprecated_cleveland/README.md b/datasets/_deprecated_cleveland/README.md index b2550eb5c..32bb9d04b 100644 --- a/datasets/_deprecated_cleveland/README.md +++ b/datasets/_deprecated_cleveland/README.md @@ -1,6 +1,6 @@ -# cleveland +# _deprecated_cleveland -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/cleveland.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_cleveland.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_cleveland/cleveland.tsv.gz b/datasets/_deprecated_cleveland/_deprecated_cleveland.tsv.gz similarity index 100% rename from datasets/_deprecated_cleveland/cleveland.tsv.gz rename to datasets/_deprecated_cleveland/_deprecated_cleveland.tsv.gz diff --git a/datasets/_deprecated_cleveland/metadata.yaml b/datasets/_deprecated_cleveland/metadata.yaml index e7e49c4e2..11457b416 100644 --- a/datasets/_deprecated_cleveland/metadata.yaml +++ b/datasets/_deprecated_cleveland/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: cleveland +dataset: _deprecated_cleveland description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -77,4 +77,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_cleveland/summary_stats.tsv b/datasets/_deprecated_cleveland/summary_stats.tsv index f54ae205f..c4a28fd15 100644 --- a/datasets/_deprecated_cleveland/summary_stats.tsv +++ b/datasets/_deprecated_cleveland/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -cleveland 303 13 3 5 5 categorical 5.0 0.19400603426679297 classification +_deprecated_cleveland 303 13 3 5 5 categorical 5.0 0.19400603426679297 classification diff --git a/datasets/_deprecated_cleveland_nominal/README.md b/datasets/_deprecated_cleveland_nominal/README.md index 946a80942..c2ac7a3f2 100644 --- a/datasets/_deprecated_cleveland_nominal/README.md +++ b/datasets/_deprecated_cleveland_nominal/README.md @@ -1,6 +1,6 @@ -# cleveland_nominal +# _deprecated_cleveland_nominal -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/cleveland_nominal.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_cleveland_nominal.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_cleveland_nominal/cleveland_nominal.tsv.gz b/datasets/_deprecated_cleveland_nominal/_deprecated_cleveland_nominal.tsv.gz similarity index 100% rename from datasets/_deprecated_cleveland_nominal/cleveland_nominal.tsv.gz rename to datasets/_deprecated_cleveland_nominal/_deprecated_cleveland_nominal.tsv.gz diff --git a/datasets/_deprecated_cleveland_nominal/metadata.yaml b/datasets/_deprecated_cleveland_nominal/metadata.yaml index a23d5f002..55a956573 100644 --- a/datasets/_deprecated_cleveland_nominal/metadata.yaml +++ b/datasets/_deprecated_cleveland_nominal/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: cleveland_nominal +dataset: _deprecated_cleveland_nominal description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -47,4 +47,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_cleveland_nominal/summary_stats.tsv b/datasets/_deprecated_cleveland_nominal/summary_stats.tsv index 5c4dcc14a..63e184501 100644 --- a/datasets/_deprecated_cleveland_nominal/summary_stats.tsv +++ b/datasets/_deprecated_cleveland_nominal/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -cleveland_nominal 303 7 3 4 0 categorical 5.0 0.19400603426679297 classification +_deprecated_cleveland_nominal 303 7 3 4 0 categorical 5.0 0.19400603426679297 classification diff --git a/datasets/_deprecated_cmc/README.md b/datasets/_deprecated_cmc/README.md index b71b45562..b73605ce8 100644 --- a/datasets/_deprecated_cmc/README.md +++ b/datasets/_deprecated_cmc/README.md @@ -1,6 +1,6 @@ -# cmc +# _deprecated_cmc -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/cmc.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_cmc.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_cmc/cmc.tsv.gz b/datasets/_deprecated_cmc/_deprecated_cmc.tsv.gz similarity index 100% rename from datasets/_deprecated_cmc/cmc.tsv.gz rename to datasets/_deprecated_cmc/_deprecated_cmc.tsv.gz diff --git a/datasets/_deprecated_cmc/metadata.yaml b/datasets/_deprecated_cmc/metadata.yaml index c555f998d..efa6a6c1d 100644 --- a/datasets/_deprecated_cmc/metadata.yaml +++ b/datasets/_deprecated_cmc/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: cmc +dataset: _deprecated_cmc description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. diff --git a/datasets/_deprecated_cmc/summary_stats.tsv b/datasets/_deprecated_cmc/summary_stats.tsv index e25d357ec..193625ad2 100644 --- a/datasets/_deprecated_cmc/summary_stats.tsv +++ b/datasets/_deprecated_cmc/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -cmc 1473 9 3 4 2 categorical 3.0 0.03070060823264104 classification +_deprecated_cmc 1473 9 3 4 2 categorical 3.0 0.03070060823264104 classification diff --git a/datasets/_deprecated_colic/README.md b/datasets/_deprecated_colic/README.md index 24f16c262..f4087b11d 100644 --- a/datasets/_deprecated_colic/README.md +++ b/datasets/_deprecated_colic/README.md @@ -1,6 +1,6 @@ -# colic +# _deprecated_colic -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/colic.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_colic.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_colic/colic.tsv.gz b/datasets/_deprecated_colic/_deprecated_colic.tsv.gz similarity index 100% rename from datasets/_deprecated_colic/colic.tsv.gz rename to datasets/_deprecated_colic/_deprecated_colic.tsv.gz diff --git a/datasets/_deprecated_colic/metadata.yaml b/datasets/_deprecated_colic/metadata.yaml index a33b69528..296b013a6 100644 --- a/datasets/_deprecated_colic/metadata.yaml +++ b/datasets/_deprecated_colic/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: colic +dataset: _deprecated_colic description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -122,4 +122,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_colic/summary_stats.tsv b/datasets/_deprecated_colic/summary_stats.tsv index 2de7c5804..c340f2b5d 100644 --- a/datasets/_deprecated_colic/summary_stats.tsv +++ b/datasets/_deprecated_colic/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -colic 368 22 1 14 7 categorical 2.0 0.0680529300567108 classification +_deprecated_colic 368 22 1 14 7 categorical 2.0 0.0680529300567108 classification diff --git a/datasets/_deprecated_contraceptive/README.md b/datasets/_deprecated_contraceptive/README.md index 4315a6c00..b577fbed2 100644 --- a/datasets/_deprecated_contraceptive/README.md +++ b/datasets/_deprecated_contraceptive/README.md @@ -1,6 +1,6 @@ -# contraceptive +# _deprecated_contraceptive -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/contraceptive.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_contraceptive.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_contraceptive/contraceptive.tsv.gz b/datasets/_deprecated_contraceptive/_deprecated_contraceptive.tsv.gz similarity index 100% rename from datasets/_deprecated_contraceptive/contraceptive.tsv.gz rename to datasets/_deprecated_contraceptive/_deprecated_contraceptive.tsv.gz diff --git a/datasets/_deprecated_contraceptive/metadata.yaml b/datasets/_deprecated_contraceptive/metadata.yaml index f1f50a07a..7e0bf84c7 100644 --- a/datasets/_deprecated_contraceptive/metadata.yaml +++ b/datasets/_deprecated_contraceptive/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: contraceptive +dataset: _deprecated_contraceptive description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -57,4 +57,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_contraceptive/summary_stats.tsv b/datasets/_deprecated_contraceptive/summary_stats.tsv index aa9bc41d4..afb02c8fa 100644 --- a/datasets/_deprecated_contraceptive/summary_stats.tsv +++ b/datasets/_deprecated_contraceptive/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -contraceptive 1473 9 3 4 2 categorical 3.0 0.03070060823264104 classification +_deprecated_contraceptive 1473 9 3 4 2 categorical 3.0 0.03070060823264104 classification diff --git a/datasets/_deprecated_credit_a/README.md b/datasets/_deprecated_credit_a/README.md index 22ded08a5..f4168a7d6 100644 --- a/datasets/_deprecated_credit_a/README.md +++ b/datasets/_deprecated_credit_a/README.md @@ -1,6 +1,6 @@ -# credit_a +# _deprecated_credit_a -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/credit_a.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_credit_a.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_credit_a/credit_a.tsv.gz b/datasets/_deprecated_credit_a/_deprecated_credit_a.tsv.gz similarity index 100% rename from datasets/_deprecated_credit_a/credit_a.tsv.gz rename to datasets/_deprecated_credit_a/_deprecated_credit_a.tsv.gz diff --git a/datasets/_deprecated_credit_a/metadata.yaml b/datasets/_deprecated_credit_a/metadata.yaml index 5da55f1b1..77f80a858 100644 --- a/datasets/_deprecated_credit_a/metadata.yaml +++ b/datasets/_deprecated_credit_a/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: credit_a +dataset: _deprecated_credit_a description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -87,4 +87,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_credit_a/summary_stats.tsv b/datasets/_deprecated_credit_a/summary_stats.tsv index 82641683c..e7e5f6d78 100644 --- a/datasets/_deprecated_credit_a/summary_stats.tsv +++ b/datasets/_deprecated_credit_a/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -credit_a 690 15 3 5 7 categorical 2.0 0.012131905061961762 classification +_deprecated_credit_a 690 15 3 5 7 categorical 2.0 0.012131905061961762 classification diff --git a/datasets/_deprecated_credit_g/README.md b/datasets/_deprecated_credit_g/README.md index 51abe9b49..efaa235bb 100644 --- a/datasets/_deprecated_credit_g/README.md +++ b/datasets/_deprecated_credit_g/README.md @@ -1,6 +1,6 @@ -# credit_g +# _deprecated_credit_g -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/credit_g.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_credit_g.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_credit_g/credit_g.tsv.gz b/datasets/_deprecated_credit_g/_deprecated_credit_g.tsv.gz similarity index 100% rename from datasets/_deprecated_credit_g/credit_g.tsv.gz rename to datasets/_deprecated_credit_g/_deprecated_credit_g.tsv.gz diff --git a/datasets/_deprecated_credit_g/metadata.yaml b/datasets/_deprecated_credit_g/metadata.yaml index 58cbc9a27..e2231ae5b 100644 --- a/datasets/_deprecated_credit_g/metadata.yaml +++ b/datasets/_deprecated_credit_g/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: credit_g +dataset: _deprecated_credit_g description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -112,4 +112,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_credit_g/summary_stats.tsv b/datasets/_deprecated_credit_g/summary_stats.tsv index f0916f9fd..6848c4b9e 100644 --- a/datasets/_deprecated_credit_g/summary_stats.tsv +++ b/datasets/_deprecated_credit_g/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -credit_g 1000 20 3 14 3 categorical 2.0 0.15999999999999998 classification +_deprecated_credit_g 1000 20 3 14 3 categorical 2.0 0.15999999999999998 classification diff --git a/datasets/_deprecated_crx/README.md b/datasets/_deprecated_crx/README.md index 03cf06e48..148582a52 100644 --- a/datasets/_deprecated_crx/README.md +++ b/datasets/_deprecated_crx/README.md @@ -1,6 +1,6 @@ -# crx +# _deprecated_crx -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/crx.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_crx.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_crx/crx.tsv.gz b/datasets/_deprecated_crx/_deprecated_crx.tsv.gz similarity index 100% rename from datasets/_deprecated_crx/crx.tsv.gz rename to datasets/_deprecated_crx/_deprecated_crx.tsv.gz diff --git a/datasets/_deprecated_crx/metadata.yaml b/datasets/_deprecated_crx/metadata.yaml index 2762c89c8..867cd1a30 100644 --- a/datasets/_deprecated_crx/metadata.yaml +++ b/datasets/_deprecated_crx/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: crx +dataset: _deprecated_crx description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -87,4 +87,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_crx/summary_stats.tsv b/datasets/_deprecated_crx/summary_stats.tsv index ce46b262e..e6d71e863 100644 --- a/datasets/_deprecated_crx/summary_stats.tsv +++ b/datasets/_deprecated_crx/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -crx 690 15 3 5 7 categorical 2.0 0.012131905061961762 classification +_deprecated_crx 690 15 3 5 7 categorical 2.0 0.012131905061961762 classification diff --git a/datasets/_deprecated_diabetes/README.md b/datasets/_deprecated_diabetes/README.md index 12f408db8..17ab0c683 100644 --- a/datasets/_deprecated_diabetes/README.md +++ b/datasets/_deprecated_diabetes/README.md @@ -1,6 +1,6 @@ -# diabetes +# _deprecated_diabetes -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/diabetes.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_diabetes.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_diabetes/diabetes.tsv.gz b/datasets/_deprecated_diabetes/_deprecated_diabetes.tsv.gz similarity index 100% rename from datasets/_deprecated_diabetes/diabetes.tsv.gz rename to datasets/_deprecated_diabetes/_deprecated_diabetes.tsv.gz diff --git a/datasets/_deprecated_diabetes/metadata.yaml b/datasets/_deprecated_diabetes/metadata.yaml index 934af14ef..4b0a916b2 100644 --- a/datasets/_deprecated_diabetes/metadata.yaml +++ b/datasets/_deprecated_diabetes/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: diabetes +dataset: _deprecated_diabetes description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -52,4 +52,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_diabetes/summary_stats.tsv b/datasets/_deprecated_diabetes/summary_stats.tsv index 969fde368..80896515d 100644 --- a/datasets/_deprecated_diabetes/summary_stats.tsv +++ b/datasets/_deprecated_diabetes/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -diabetes 768 8 0 0 8 categorical 2.0 0.09125434027777776 classification +_deprecated_diabetes 768 8 0 0 8 categorical 2.0 0.09125434027777776 classification diff --git a/datasets/_deprecated_flare/README.md b/datasets/_deprecated_flare/README.md index 90f2c849e..89decf0e9 100644 --- a/datasets/_deprecated_flare/README.md +++ b/datasets/_deprecated_flare/README.md @@ -1,6 +1,6 @@ -# flare +# _deprecated_flare -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/flare.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_flare.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_flare/flare.tsv.gz b/datasets/_deprecated_flare/_deprecated_flare.tsv.gz similarity index 100% rename from datasets/_deprecated_flare/flare.tsv.gz rename to datasets/_deprecated_flare/_deprecated_flare.tsv.gz diff --git a/datasets/_deprecated_flare/metadata.yaml b/datasets/_deprecated_flare/metadata.yaml index a1d4cb09e..fd5a38ab1 100644 --- a/datasets/_deprecated_flare/metadata.yaml +++ b/datasets/_deprecated_flare/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: flare +dataset: _deprecated_flare description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -62,4 +62,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_flare/summary_stats.tsv b/datasets/_deprecated_flare/summary_stats.tsv index 6e6e7bcde..2308ea84b 100644 --- a/datasets/_deprecated_flare/summary_stats.tsv +++ b/datasets/_deprecated_flare/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -flare 1066 10 4 6 0 categorical 2.0 0.43367043426531815 classification +_deprecated_flare 1066 10 4 6 0 categorical 2.0 0.43367043426531815 classification diff --git a/datasets/_deprecated_german/README.md b/datasets/_deprecated_german/README.md index 76ab506b1..1adfed884 100644 --- a/datasets/_deprecated_german/README.md +++ b/datasets/_deprecated_german/README.md @@ -1,6 +1,6 @@ -# german +# _deprecated_german -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/german.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_german.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_german/german.tsv.gz b/datasets/_deprecated_german/_deprecated_german.tsv.gz similarity index 100% rename from datasets/_deprecated_german/german.tsv.gz rename to datasets/_deprecated_german/_deprecated_german.tsv.gz diff --git a/datasets/_deprecated_german/metadata.yaml b/datasets/_deprecated_german/metadata.yaml index e81f436e3..79f650505 100644 --- a/datasets/_deprecated_german/metadata.yaml +++ b/datasets/_deprecated_german/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: german +dataset: _deprecated_german description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -112,4 +112,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_german/summary_stats.tsv b/datasets/_deprecated_german/summary_stats.tsv index f29c5a3e7..afcaf8ba2 100644 --- a/datasets/_deprecated_german/summary_stats.tsv +++ b/datasets/_deprecated_german/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -german 1000 20 3 14 3 categorical 2.0 0.15999999999999998 classification +_deprecated_german 1000 20 3 14 3 categorical 2.0 0.15999999999999998 classification diff --git a/datasets/_deprecated_glass/README.md b/datasets/_deprecated_glass/README.md index ca363c401..c376f8b9c 100644 --- a/datasets/_deprecated_glass/README.md +++ b/datasets/_deprecated_glass/README.md @@ -1,6 +1,6 @@ -# glass +# _deprecated_glass -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/glass.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_glass.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_glass/glass.tsv.gz b/datasets/_deprecated_glass/_deprecated_glass.tsv.gz similarity index 100% rename from datasets/_deprecated_glass/glass.tsv.gz rename to datasets/_deprecated_glass/_deprecated_glass.tsv.gz diff --git a/datasets/_deprecated_glass/metadata.yaml b/datasets/_deprecated_glass/metadata.yaml index e9e745e66..70f64ab59 100644 --- a/datasets/_deprecated_glass/metadata.yaml +++ b/datasets/_deprecated_glass/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: glass +dataset: _deprecated_glass description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -57,4 +57,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_glass/summary_stats.tsv b/datasets/_deprecated_glass/summary_stats.tsv index 263b81bbd..4ce9173d7 100644 --- a/datasets/_deprecated_glass/summary_stats.tsv +++ b/datasets/_deprecated_glass/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -glass 205 9 0 0 9 categorical 5.0 0.10618679357525282 classification +_deprecated_glass 205 9 0 0 9 categorical 5.0 0.10618679357525282 classification diff --git a/datasets/_deprecated_heart_c/README.md b/datasets/_deprecated_heart_c/README.md index 3838332f6..fcf3a75c6 100644 --- a/datasets/_deprecated_heart_c/README.md +++ b/datasets/_deprecated_heart_c/README.md @@ -1,6 +1,6 @@ -# heart_c +# _deprecated_heart_c -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/heart_c.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_heart_c.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_heart_c/heart_c.tsv.gz b/datasets/_deprecated_heart_c/_deprecated_heart_c.tsv.gz similarity index 100% rename from datasets/_deprecated_heart_c/heart_c.tsv.gz rename to datasets/_deprecated_heart_c/_deprecated_heart_c.tsv.gz diff --git a/datasets/_deprecated_heart_c/metadata.yaml b/datasets/_deprecated_heart_c/metadata.yaml index fb2054d16..e9b25f8c5 100644 --- a/datasets/_deprecated_heart_c/metadata.yaml +++ b/datasets/_deprecated_heart_c/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: heart_c +dataset: _deprecated_heart_c description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -77,4 +77,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_heart_c/summary_stats.tsv b/datasets/_deprecated_heart_c/summary_stats.tsv index b6edab1ab..8aaa90902 100644 --- a/datasets/_deprecated_heart_c/summary_stats.tsv +++ b/datasets/_deprecated_heart_c/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -heart_c 303 13 3 5 5 categorical 2.0 0.007940398000196063 classification +_deprecated_heart_c 303 13 3 5 5 categorical 2.0 0.007940398000196063 classification diff --git a/datasets/_deprecated_heart_h/README.md b/datasets/_deprecated_heart_h/README.md index 56208dcea..49108ec9c 100644 --- a/datasets/_deprecated_heart_h/README.md +++ b/datasets/_deprecated_heart_h/README.md @@ -1,6 +1,6 @@ -# heart_h +# _deprecated_heart_h -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/heart_h.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_heart_h.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_heart_h/heart_h.tsv.gz b/datasets/_deprecated_heart_h/_deprecated_heart_h.tsv.gz similarity index 100% rename from datasets/_deprecated_heart_h/heart_h.tsv.gz rename to datasets/_deprecated_heart_h/_deprecated_heart_h.tsv.gz diff --git a/datasets/_deprecated_heart_h/metadata.yaml b/datasets/_deprecated_heart_h/metadata.yaml index 38a7f46e4..2ef488c7e 100644 --- a/datasets/_deprecated_heart_h/metadata.yaml +++ b/datasets/_deprecated_heart_h/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: heart_h +dataset: _deprecated_heart_h description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -77,4 +77,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_heart_h/summary_stats.tsv b/datasets/_deprecated_heart_h/summary_stats.tsv index 9c8953da3..ebd010e5f 100644 --- a/datasets/_deprecated_heart_h/summary_stats.tsv +++ b/datasets/_deprecated_heart_h/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -heart_h 294 13 2 6 5 categorical 2.0 0.07779166088203991 classification +_deprecated_heart_h 294 13 2 6 5 categorical 2.0 0.07779166088203991 classification diff --git a/datasets/_deprecated_heart_statlog/README.md b/datasets/_deprecated_heart_statlog/README.md index 246a98762..faa50392e 100644 --- a/datasets/_deprecated_heart_statlog/README.md +++ b/datasets/_deprecated_heart_statlog/README.md @@ -1,6 +1,6 @@ -# heart_statlog +# _deprecated_heart_statlog -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/heart_statlog.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_heart_statlog.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_heart_statlog/heart_statlog.tsv.gz b/datasets/_deprecated_heart_statlog/_deprecated_heart_statlog.tsv.gz similarity index 100% rename from datasets/_deprecated_heart_statlog/heart_statlog.tsv.gz rename to datasets/_deprecated_heart_statlog/_deprecated_heart_statlog.tsv.gz diff --git a/datasets/_deprecated_heart_statlog/metadata.yaml b/datasets/_deprecated_heart_statlog/metadata.yaml index 415c5ffb2..7f1dd31d1 100644 --- a/datasets/_deprecated_heart_statlog/metadata.yaml +++ b/datasets/_deprecated_heart_statlog/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: heart_statlog +dataset: _deprecated_heart_statlog description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -77,4 +77,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_heart_statlog/summary_stats.tsv b/datasets/_deprecated_heart_statlog/summary_stats.tsv index 35a010096..02529670b 100644 --- a/datasets/_deprecated_heart_statlog/summary_stats.tsv +++ b/datasets/_deprecated_heart_statlog/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -heart_statlog 270 13 3 5 5 categorical 2.0 0.01234567901234569 classification +_deprecated_heart_statlog 270 13 3 5 5 categorical 2.0 0.01234567901234569 classification diff --git a/datasets/_deprecated_horse_colic/README.md b/datasets/_deprecated_horse_colic/README.md index 04c44cefb..9bf84868f 100644 --- a/datasets/_deprecated_horse_colic/README.md +++ b/datasets/_deprecated_horse_colic/README.md @@ -1,6 +1,6 @@ -# horse_colic +# _deprecated_horse_colic -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/horse_colic.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_horse_colic.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_horse_colic/horse_colic.tsv.gz b/datasets/_deprecated_horse_colic/_deprecated_horse_colic.tsv.gz similarity index 100% rename from datasets/_deprecated_horse_colic/horse_colic.tsv.gz rename to datasets/_deprecated_horse_colic/_deprecated_horse_colic.tsv.gz diff --git a/datasets/_deprecated_horse_colic/metadata.yaml b/datasets/_deprecated_horse_colic/metadata.yaml index 6060351c8..d815a5849 100644 --- a/datasets/_deprecated_horse_colic/metadata.yaml +++ b/datasets/_deprecated_horse_colic/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: horse_colic +dataset: _deprecated_horse_colic description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -122,4 +122,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_horse_colic/summary_stats.tsv b/datasets/_deprecated_horse_colic/summary_stats.tsv index e095b469f..b58c08b0e 100644 --- a/datasets/_deprecated_horse_colic/summary_stats.tsv +++ b/datasets/_deprecated_horse_colic/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -horse_colic 368 22 1 14 7 categorical 2.0 0.0680529300567108 classification +_deprecated_horse_colic 368 22 1 14 7 categorical 2.0 0.0680529300567108 classification diff --git a/datasets/_deprecated_house_votes_84/README.md b/datasets/_deprecated_house_votes_84/README.md index dfc0c8842..b4c166298 100644 --- a/datasets/_deprecated_house_votes_84/README.md +++ b/datasets/_deprecated_house_votes_84/README.md @@ -1,6 +1,6 @@ -# house_votes_84 +# _deprecated_house_votes_84 -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/house_votes_84.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_house_votes_84.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_house_votes_84/house_votes_84.tsv.gz b/datasets/_deprecated_house_votes_84/_deprecated_house_votes_84.tsv.gz similarity index 100% rename from datasets/_deprecated_house_votes_84/house_votes_84.tsv.gz rename to datasets/_deprecated_house_votes_84/_deprecated_house_votes_84.tsv.gz diff --git a/datasets/_deprecated_house_votes_84/metadata.yaml b/datasets/_deprecated_house_votes_84/metadata.yaml index 703b60d56..3644d4272 100644 --- a/datasets/_deprecated_house_votes_84/metadata.yaml +++ b/datasets/_deprecated_house_votes_84/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: house_votes_84 +dataset: _deprecated_house_votes_84 description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -92,4 +92,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_house_votes_84/summary_stats.tsv b/datasets/_deprecated_house_votes_84/summary_stats.tsv index 96050ce26..6b0cd30b9 100644 --- a/datasets/_deprecated_house_votes_84/summary_stats.tsv +++ b/datasets/_deprecated_house_votes_84/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -house_votes_84 435 16 0 16 0 categorical 2.0 0.05179548156956005 classification +_deprecated_house_votes_84 435 16 0 16 0 categorical 2.0 0.05179548156956005 classification diff --git a/datasets/_deprecated_hungarian/README.md b/datasets/_deprecated_hungarian/README.md index c40434472..c29f4ae9b 100644 --- a/datasets/_deprecated_hungarian/README.md +++ b/datasets/_deprecated_hungarian/README.md @@ -1,6 +1,6 @@ -# hungarian +# _deprecated_hungarian -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/hungarian.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_hungarian.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_hungarian/hungarian.tsv.gz b/datasets/_deprecated_hungarian/_deprecated_hungarian.tsv.gz similarity index 100% rename from datasets/_deprecated_hungarian/hungarian.tsv.gz rename to datasets/_deprecated_hungarian/_deprecated_hungarian.tsv.gz diff --git a/datasets/_deprecated_hungarian/metadata.yaml b/datasets/_deprecated_hungarian/metadata.yaml index 807467f5c..6dadaf77d 100644 --- a/datasets/_deprecated_hungarian/metadata.yaml +++ b/datasets/_deprecated_hungarian/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: hungarian +dataset: _deprecated_hungarian description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -77,4 +77,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_hungarian/summary_stats.tsv b/datasets/_deprecated_hungarian/summary_stats.tsv index a3637bcbe..ced8153cd 100644 --- a/datasets/_deprecated_hungarian/summary_stats.tsv +++ b/datasets/_deprecated_hungarian/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -hungarian 294 13 2 6 5 categorical 2.0 0.07779166088203991 classification +_deprecated_hungarian 294 13 2 6 5 categorical 2.0 0.07779166088203991 classification diff --git a/datasets/_deprecated_pima/README.md b/datasets/_deprecated_pima/README.md index e29d50e5d..f139c9691 100644 --- a/datasets/_deprecated_pima/README.md +++ b/datasets/_deprecated_pima/README.md @@ -1,6 +1,6 @@ -# pima +# _deprecated_pima -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/pima.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_pima.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_pima/pima.tsv.gz b/datasets/_deprecated_pima/_deprecated_pima.tsv.gz similarity index 100% rename from datasets/_deprecated_pima/pima.tsv.gz rename to datasets/_deprecated_pima/_deprecated_pima.tsv.gz diff --git a/datasets/_deprecated_pima/metadata.yaml b/datasets/_deprecated_pima/metadata.yaml index 31915c390..19c13ed4c 100644 --- a/datasets/_deprecated_pima/metadata.yaml +++ b/datasets/_deprecated_pima/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: pima +dataset: _deprecated_pima description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -52,4 +52,3 @@ features: description: code: transform: -deprecated: true \ No newline at end of file diff --git a/datasets/_deprecated_pima/summary_stats.tsv b/datasets/_deprecated_pima/summary_stats.tsv index a6405742a..fdc3b1bf3 100644 --- a/datasets/_deprecated_pima/summary_stats.tsv +++ b/datasets/_deprecated_pima/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -pima 768 8 0 0 8 categorical 2.0 0.09125434027777776 classification +_deprecated_pima 768 8 0 0 8 categorical 2.0 0.09125434027777776 classification diff --git a/datasets/_deprecated_prnn_fglass/README.md b/datasets/_deprecated_prnn_fglass/README.md index 5d61060e1..31840b863 100644 --- a/datasets/_deprecated_prnn_fglass/README.md +++ b/datasets/_deprecated_prnn_fglass/README.md @@ -1,6 +1,6 @@ -# prnn_fglass +# _deprecated_prnn_fglass -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/prnn_fglass.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_prnn_fglass.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_prnn_fglass/prnn_fglass.tsv.gz b/datasets/_deprecated_prnn_fglass/_deprecated_prnn_fglass.tsv.gz similarity index 100% rename from datasets/_deprecated_prnn_fglass/prnn_fglass.tsv.gz rename to datasets/_deprecated_prnn_fglass/_deprecated_prnn_fglass.tsv.gz diff --git a/datasets/_deprecated_prnn_fglass/metadata.yaml b/datasets/_deprecated_prnn_fglass/metadata.yaml index aff32fd37..b32079721 100644 --- a/datasets/_deprecated_prnn_fglass/metadata.yaml +++ b/datasets/_deprecated_prnn_fglass/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: prnn_fglass +dataset: _deprecated_prnn_fglass description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -57,4 +57,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_prnn_fglass/summary_stats.tsv b/datasets/_deprecated_prnn_fglass/summary_stats.tsv index 662567afc..610574b59 100644 --- a/datasets/_deprecated_prnn_fglass/summary_stats.tsv +++ b/datasets/_deprecated_prnn_fglass/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -prnn_fglass 205 9 0 0 9 categorical 5.0 0.10618679357525282 classification +_deprecated_prnn_fglass 205 9 0 0 9 categorical 5.0 0.10618679357525282 classification diff --git a/datasets/_deprecated_solar_flare_1/README.md b/datasets/_deprecated_solar_flare_1/README.md index 4f658feb9..e0fe8b823 100644 --- a/datasets/_deprecated_solar_flare_1/README.md +++ b/datasets/_deprecated_solar_flare_1/README.md @@ -1,6 +1,6 @@ -# solar_flare_1 +# _deprecated_solar_flare_1 -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/solar_flare_1.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_solar_flare_1.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_solar_flare_1/solar_flare_1.tsv.gz b/datasets/_deprecated_solar_flare_1/_deprecated_solar_flare_1.tsv.gz similarity index 100% rename from datasets/_deprecated_solar_flare_1/solar_flare_1.tsv.gz rename to datasets/_deprecated_solar_flare_1/_deprecated_solar_flare_1.tsv.gz diff --git a/datasets/_deprecated_solar_flare_1/metadata.yaml b/datasets/_deprecated_solar_flare_1/metadata.yaml index f680a0a8f..3964729d5 100644 --- a/datasets/_deprecated_solar_flare_1/metadata.yaml +++ b/datasets/_deprecated_solar_flare_1/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: solar_flare_1 +dataset: _deprecated_solar_flare_1 description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -72,4 +72,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_solar_flare_1/summary_stats.tsv b/datasets/_deprecated_solar_flare_1/summary_stats.tsv index f35242775..e627b4703 100644 --- a/datasets/_deprecated_solar_flare_1/summary_stats.tsv +++ b/datasets/_deprecated_solar_flare_1/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -solar_flare_1 315 12 7 5 0 categorical 5.0 0.03232552280171328 classification +_deprecated_solar_flare_1 315 12 7 5 0 categorical 5.0 0.03232552280171328 classification diff --git a/datasets/_deprecated_solar_flare_2/README.md b/datasets/_deprecated_solar_flare_2/README.md index b50d6efe4..5f8908cdb 100644 --- a/datasets/_deprecated_solar_flare_2/README.md +++ b/datasets/_deprecated_solar_flare_2/README.md @@ -1,6 +1,6 @@ -# solar_flare_2 +# _deprecated_solar_flare_2 -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/solar_flare_2.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_solar_flare_2.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_solar_flare_2/solar_flare_2.tsv.gz b/datasets/_deprecated_solar_flare_2/_deprecated_solar_flare_2.tsv.gz similarity index 100% rename from datasets/_deprecated_solar_flare_2/solar_flare_2.tsv.gz rename to datasets/_deprecated_solar_flare_2/_deprecated_solar_flare_2.tsv.gz diff --git a/datasets/_deprecated_solar_flare_2/metadata.yaml b/datasets/_deprecated_solar_flare_2/metadata.yaml index 6423b419e..506582eb1 100644 --- a/datasets/_deprecated_solar_flare_2/metadata.yaml +++ b/datasets/_deprecated_solar_flare_2/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: solar_flare_2 +dataset: _deprecated_solar_flare_2 description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -72,4 +72,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_solar_flare_2/summary_stats.tsv b/datasets/_deprecated_solar_flare_2/summary_stats.tsv index 677a81f39..7e56b5ac9 100644 --- a/datasets/_deprecated_solar_flare_2/summary_stats.tsv +++ b/datasets/_deprecated_solar_flare_2/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -solar_flare_2 1066 12 4 8 0 categorical 6.0 0.057334145285456294 classification +_deprecated_solar_flare_2 1066 12 4 8 0 categorical 6.0 0.057334145285456294 classification diff --git a/datasets/_deprecated_vote/README.md b/datasets/_deprecated_vote/README.md index f2fd29ea6..7ea77ab13 100644 --- a/datasets/_deprecated_vote/README.md +++ b/datasets/_deprecated_vote/README.md @@ -1,6 +1,6 @@ -# vote +# _deprecated_vote -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/vote.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_vote.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_vote/vote.tsv.gz b/datasets/_deprecated_vote/_deprecated_vote.tsv.gz similarity index 100% rename from datasets/_deprecated_vote/vote.tsv.gz rename to datasets/_deprecated_vote/_deprecated_vote.tsv.gz diff --git a/datasets/_deprecated_vote/metadata.yaml b/datasets/_deprecated_vote/metadata.yaml index cb89de7f3..ba53068ed 100644 --- a/datasets/_deprecated_vote/metadata.yaml +++ b/datasets/_deprecated_vote/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: vote +dataset: _deprecated_vote description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -92,4 +92,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_vote/summary_stats.tsv b/datasets/_deprecated_vote/summary_stats.tsv index cea5bffa0..80cc806d6 100644 --- a/datasets/_deprecated_vote/summary_stats.tsv +++ b/datasets/_deprecated_vote/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -vote 435 16 0 16 0 categorical 2.0 0.05179548156956005 classification +_deprecated_vote 435 16 0 16 0 categorical 2.0 0.05179548156956005 classification diff --git a/datasets/_deprecated_wdbc/README.md b/datasets/_deprecated_wdbc/README.md index 5dc9f5eaf..97cba7d66 100644 --- a/datasets/_deprecated_wdbc/README.md +++ b/datasets/_deprecated_wdbc/README.md @@ -1,6 +1,6 @@ -# wdbc +# _deprecated_wdbc -[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/wdbc.html) +[**Pandas Profiling Report**](https://epistasislab.github.io/pmlb/profile/_deprecated_wdbc.html) [Metadata](metadata.yaml) | [Summary Statistics](summary_stats.tsv) diff --git a/datasets/_deprecated_wdbc/wdbc.tsv.gz b/datasets/_deprecated_wdbc/_deprecated_wdbc.tsv.gz similarity index 100% rename from datasets/_deprecated_wdbc/wdbc.tsv.gz rename to datasets/_deprecated_wdbc/_deprecated_wdbc.tsv.gz diff --git a/datasets/_deprecated_wdbc/metadata.yaml b/datasets/_deprecated_wdbc/metadata.yaml index 11ef10ecc..44be8c7ac 100644 --- a/datasets/_deprecated_wdbc/metadata.yaml +++ b/datasets/_deprecated_wdbc/metadata.yaml @@ -1,5 +1,5 @@ # Reviewed by [your name here] -dataset: wdbc +dataset: _deprecated_wdbc description: None yet. See our contributing guide to help us add one. source: None yet. See our contributing guide to help us add one. publication: None yet. See our contributing guide to help us add one. @@ -162,4 +162,3 @@ features: description: code: transform: -deprecated: true diff --git a/datasets/_deprecated_wdbc/summary_stats.tsv b/datasets/_deprecated_wdbc/summary_stats.tsv index ebe60834a..6110e2385 100644 --- a/datasets/_deprecated_wdbc/summary_stats.tsv +++ b/datasets/_deprecated_wdbc/summary_stats.tsv @@ -1,2 +1,2 @@ dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features endpoint_type n_classes imbalance task -wdbc 569 30 0 0 30 categorical 2.0 0.06493987849061501 classification +_deprecated_wdbc 569 30 0 0 30 categorical 2.0 0.06493987849061501 classification diff --git a/datasets/breast_cancer_wisconsin_prognostic/summary_stats.tsv b/datasets/breast_cancer_wisconsin_prognostic/summary_stats.tsv deleted file mode 100644 index c6f859362..000000000 --- a/datasets/breast_cancer_wisconsin_prognostic/summary_stats.tsv +++ /dev/null @@ -1,2 +0,0 @@ -dataset n_instances n_features n_binary_features n_categorical_features n_continuous_features n_classes imbalance task -breast_cancer_wisconsin_prognostic 198 32 0 0 32 1 survival diff --git a/docs_sources/index.Rmd b/docs_sources/index.Rmd index 18335dda6..1b444e912 100644 --- a/docs_sources/index.Rmd +++ b/docs_sources/index.Rmd @@ -44,7 +44,7 @@ ggp <- ply_dat %>% scale_x_log10(name = 'Number of observations', ) + scale_y_log10(name = 'Number of features') + - scale_color_manual(values = c('#ca446f', '#009392')) + + scale_color_manual(values = c('#ca446f', '#009392', 'grey40')) + NULL ply <- plotly_build(ggp) diff --git a/docs_sources/using-python.Rmd b/docs_sources/using-python.Rmd index be22bb8fd..10c81e838 100644 --- a/docs_sources/using-python.Rmd +++ b/docs_sources/using-python.Rmd @@ -90,6 +90,10 @@ from pmlb import fetch_data, classification_dataset_names logit_test_scores = [] gnb_test_scores = [] +classification_dataset_names = [ + name for name in classification_dataset_names if not name.startswith("_deprecated_") +] + for classification_dataset in classification_dataset_names[:40]: # Read in the datasets and split them into training/testing X, y = fetch_data(classification_dataset, return_X_y=True) diff --git a/pmlb/all_summary_stats.tsv b/pmlb/all_summary_stats.tsv index 9011bd805..8e99941a4 100644 --- a/pmlb/all_summary_stats.tsv +++ b/pmlb/all_summary_stats.tsv @@ -13,10 +13,8 @@ dataset n_instances n_features n_binary_features n_categorical_features n_contin 1203_BNG_pwLinear 177147 10 1 9 0 continuous 176025.0 3.5972484453226156e-08 regression 1595_poker 1025010 10 0 5 5 continuous 10.0 0.3693735838721754 regression 192_vineyard 52 2 0 0 2 continuous 19.0 0.0404750164365548 regression -195_auto_price 159 15 0 1 14 continuous 145.0 0.0005037819354895 regression 197_cpu_act 8192 21 0 0 21 continuous 56.0 0.0205723459070379 regression 201_pol 15000 48 0 22 26 continuous 11.0 0.3746815528888889 regression -207_autoPrice 159 15 0 1 14 continuous 145.0 0.0005037819354895 regression 210_cloud 108 5 1 1 3 continuous 94.0 0.0013791170700768 regression 215_2dplanes 40768 10 1 9 0 continuous 40368.0 2.419008481400585e-07 regression 218_house_8L 22784 8 0 0 8 continuous 2045.0 0.0193534322439784 regression @@ -127,6 +125,41 @@ GAMETES_Heterogeneity_20atts_1600_Het_0.4_0.2_50_EDM_2_001 1600 20 1 19 0 catego GAMETES_Heterogeneity_20atts_1600_Het_0.4_0.2_75_EDM_2_001 1600 20 1 19 0 categorical 2.0 0.0 classification Hill_Valley_with_noise 1212 100 0 0 100 categorical 2.0 0.0 classification Hill_Valley_without_noise 1212 100 0 0 100 categorical 2.0 9.802960494069226e-05 classification +_deprecated_195_auto_price 159 15 0 1 14 continuous 145.0 0.0005037819354895 regression +_deprecated_207_autoPrice 159 15 0 1 14 continuous 145.0 0.0005037819354895 regression +_deprecated_australian 690 14 4 3 7 categorical 2.0 0.0121319050619617 classification +_deprecated_auto 202 25 3 6 16 categorical 5.0 0.0453754533869228 classification +_deprecated_breast 699 10 0 8 2 categorical 2.0 0.0963751609186227 classification +_deprecated_breast_cancer_wisconsin 569 30 0 0 30 categorical 2.0 0.064939878490615 classification +_deprecated_breast_w 699 9 0 8 1 categorical 2.0 0.0963751609186227 classification +_deprecated_buggyCrx 690 15 3 5 7 categorical 2.0 0.0121319050619617 classification +_deprecated_car 1728 6 0 6 0 categorical 4.0 0.3902883159007773 classification +_deprecated_car_evaluation 1728 21 21 0 0 categorical 4.0 0.3902883159007773 classification +_deprecated_cleve 303 13 3 5 5 categorical 2.0 0.007940398000196 classification +_deprecated_cleveland 303 13 3 5 5 categorical 5.0 0.1940060342667929 classification +_deprecated_cleveland_nominal 303 7 3 4 0 categorical 5.0 0.1940060342667929 classification +_deprecated_cmc 1473 9 3 4 2 categorical 3.0 0.030700608232641 classification +_deprecated_colic 368 22 1 14 7 categorical 2.0 0.0680529300567108 classification +_deprecated_contraceptive 1473 9 3 4 2 categorical 3.0 0.030700608232641 classification +_deprecated_credit_a 690 15 3 5 7 categorical 2.0 0.0121319050619617 classification +_deprecated_credit_g 1000 20 3 14 3 categorical 2.0 0.1599999999999999 classification +_deprecated_crx 690 15 3 5 7 categorical 2.0 0.0121319050619617 classification +_deprecated_diabetes 768 8 0 0 8 categorical 2.0 0.0912543402777777 classification +_deprecated_flare 1066 10 4 6 0 categorical 2.0 0.4336704342653181 classification +_deprecated_german 1000 20 3 14 3 categorical 2.0 0.1599999999999999 classification +_deprecated_glass 205 9 0 0 9 categorical 5.0 0.1061867935752528 classification +_deprecated_heart_c 303 13 3 5 5 categorical 2.0 0.007940398000196 classification +_deprecated_heart_h 294 13 2 6 5 categorical 2.0 0.0777916608820399 classification +_deprecated_heart_statlog 270 13 3 5 5 categorical 2.0 0.0123456790123456 classification +_deprecated_horse_colic 368 22 1 14 7 categorical 2.0 0.0680529300567108 classification +_deprecated_house_votes_84 435 16 0 16 0 categorical 2.0 0.05179548156956 classification +_deprecated_hungarian 294 13 2 6 5 categorical 2.0 0.0777916608820399 classification +_deprecated_pima 768 8 0 0 8 categorical 2.0 0.0912543402777777 classification +_deprecated_prnn_fglass 205 9 0 0 9 categorical 5.0 0.1061867935752528 classification +_deprecated_solar_flare_1 315 12 7 5 0 categorical 5.0 0.0323255228017132 classification +_deprecated_solar_flare_2 1066 12 4 8 0 categorical 6.0 0.0573341452854562 classification +_deprecated_vote 435 16 0 16 0 categorical 2.0 0.05179548156956 classification +_deprecated_wdbc 569 30 0 0 30 categorical 2.0 0.064939878490615 classification adult 48842 14 1 4 9 categorical 2.0 0.2718959907996784 classification agaricus_lepiota 8145 22 5 16 1 categorical 2.0 0.0014579364729441 classification allbp 3772 29 19 4 6 categorical 3.0 0.8755228428707819 classification @@ -150,43 +183,35 @@ analcatdata_japansolvent 52 9 0 0 9 categorical 2.0 0.0014792899408284 classific analcatdata_lawsuit 264 4 1 0 3 categorical 2.0 0.7328397612488521 classification ann_thyroid 7200 21 15 0 6 categorical 3.0 0.7904668981481481 classification appendicitis 106 7 0 0 7 categorical 2.0 0.3645425418298327 classification -australian 690 14 4 3 7 categorical 2.0 0.0121319050619617 classification -auto 202 25 3 6 16 categorical 5.0 0.0453754533869228 classification +auto_insurance_losses 164 24 3 7 14 continuous 51.0 0.0084428911362284 regression +auto_insurance_price 201 23 4 6 13 continuous 186.0 0.0003431812297929 regression +auto_insurance_symboling 205 24 4 6 14 ordinal 6.0 0.0755788221296847 classification backache 180 32 22 4 6 categorical 2.0 0.521604938271605 classification balance_scale 625 4 0 4 0 categorical 3.0 0.14622976 classification banana 5300 2 0 0 2 categorical 2.0 0.0106907796368814 classification biomed 209 8 0 1 7 categorical 2.0 0.079691399006433 classification -breast 699 10 0 8 2 categorical 2.0 0.0963751609186227 classification breast_cancer 286 9 2 6 1 categorical 2.0 0.1645068218494792 classification -breast_cancer_wisconsin 569 30 0 0 30 categorical 2.0 0.064939878490615 classification -breast_w 699 9 0 8 1 categorical 2.0 0.0963751609186227 classification -buggyCrx 690 15 3 5 7 categorical 2.0 0.0121319050619617 classification +breast_cancer_wisconsin_diagnostic 569 30 0 0 30 binary 2.0 0.064939878490615 classification +breast_cancer_wisconsin_original 699 9 0 0 9 binary 2.0 0.0963751609186227 classification bupa 345 5 0 0 5 categorical 2.0 0.0004116782188615 classification calendarDOW 399 32 2 18 12 categorical 5.0 0.0140294344884768 classification -car 1728 6 0 6 0 categorical 4.0 0.3902883159007773 classification -car_evaluation 1728 21 21 0 0 categorical 4.0 0.3902883159007773 classification +car_evaluation 1728 6 0 6 0 categorical 4.0 0.390288315900777 classification cars 392 8 0 1 7 categorical 3.0 0.1919968242399 classification chess 3196 36 35 1 0 categorical 2.0 0.0019740727223171 classification churn 5000 20 2 2 16 categorical 2.0 0.5143758400000001 classification clean1 476 168 0 0 168 categorical 2.0 0.016965609773321 classification clean2 6598 168 0 0 168 categorical 2.0 0.478483150625434 classification -cleve 303 13 3 5 5 categorical 2.0 0.007940398000196 classification -cleveland 303 13 3 5 5 categorical 5.0 0.1940060342667929 classification -cleveland_nominal 303 7 3 4 0 categorical 5.0 0.1940060342667929 classification cloud 108 7 1 0 6 categorical 4.0 0.0050297210791037 classification -cmc 1473 9 3 4 2 categorical 3.0 0.030700608232641 classification coil2000 9822 85 5 79 1 categorical 2.0 0.7755902757953309 classification -colic 368 22 1 14 7 categorical 2.0 0.0680529300567108 classification collins 485 23 0 1 22 categorical 13.0 0.0219810819428207 classification confidence 72 3 0 0 3 categorical 6.0 0.0 classification +congressional_voting_records 435 16 0 16 0 binary 2.0 0.0517954815695601 classification connect_4 67557 42 0 42 0 categorical 3.0 0.2546625519372754 classification -contraceptive 1473 9 3 4 2 categorical 3.0 0.030700608232641 classification +contraceptive_method 1473 9 3 4 2 categorical 3.0 0.030700608232641 classification corral 160 6 6 0 0 categorical 2.0 0.015625 classification -credit_a 690 15 3 5 7 categorical 2.0 0.0121319050619617 classification -credit_g 1000 20 3 14 3 categorical 2.0 0.1599999999999999 classification -crx 690 15 3 5 7 categorical 2.0 0.0121319050619617 classification +credit_approval_australia 690 15 4 5 6 binary 2.0 0.0121319050619618 classification +credit_approval_germany 1000 20 3 14 3 binary 2.0 0.16 classification dermatology 366 34 1 32 1 categorical 6.0 0.0414583893218668 classification -diabetes 768 8 0 0 8 categorical 2.0 0.0912543402777777 classification dis 3772 29 19 4 6 categorical 2.0 0.9394399094067016 classification dna 3186 180 180 0 0 categorical 3.0 0.0776847980158012 classification ecoli 327 7 0 0 7 categorical 5.0 0.1089648271282813 classification @@ -311,19 +336,17 @@ feynman_test_7 100000 5 0 0 5 continuous 100000.0 0.0 regression feynman_test_8 100000 4 0 0 4 continuous 100000.0 0.0 regression feynman_test_9 100000 5 0 0 5 continuous 100000.0 0.0 regression flags 178 43 36 5 2 categorical 5.0 0.04391806590077 classification -flare 1066 10 4 6 0 categorical 2.0 0.4336704342653181 classification -german 1000 20 3 14 3 categorical 2.0 0.1599999999999999 classification -glass 205 9 0 0 9 categorical 5.0 0.1061867935752528 classification glass2 163 9 0 0 9 categorical 2.0 0.0045541796830893 classification haberman 306 3 0 0 3 categorical 2.0 0.2214532871972319 classification hayes_roth 160 4 0 4 0 categorical 3.0 0.0438671874999999 classification -heart_c 303 13 3 5 5 categorical 2.0 0.007940398000196 classification -heart_h 294 13 2 6 5 categorical 2.0 0.0777916608820399 classification -heart_statlog 270 13 3 5 5 categorical 2.0 0.0123456790123456 classification +heart_disease_cleveland 303 13 3 4 6 binary 2.0 0.0068076114542147 classification +heart_disease_hungarian 294 13 3 4 6 binary 2.0 0.0777916608820399 classification +heart_disease_va_long_beach 200 13 3 4 6 binary 2.0 0.2401 classification +heart_disease_zurich 123 12 3 4 5 binary 2.0 0.756758543195188 classification hepatitis 155 19 3 10 6 categorical 2.0 0.3446826222684704 classification -horse_colic 368 22 1 14 7 categorical 2.0 0.0680529300567108 classification -house_votes_84 435 16 0 16 0 categorical 2.0 0.05179548156956 classification -hungarian 294 13 2 6 5 categorical 2.0 0.0777916608820399 classification +horse_colic_lesion_type 368 20 2 11 7 categorical 5.0 0.113506468572779 classification +horse_colic_outcome 366 21 3 11 7 categorical 3.0 0.185859535967034 classification +horse_colic_surgery 368 20 2 11 7 binary 2.0 0.0680529300567108 classification hypothyroid 3163 25 17 1 7 categorical 2.0 0.8181582834802777 classification ionosphere 351 34 1 1 32 categorical 2.0 0.0795529257067718 classification iris 150 4 0 0 4 categorical 3.0 0.0 classification @@ -364,11 +387,9 @@ parity5+5 1124 10 10 0 0 categorical 2.0 7.915299958207264e-05 classification pendigits 10992 16 0 0 16 categorical 10.0 0.0001780697519917 classification penguins 344 7 1 2 4 categorical 3.0 0.0463764196863169 classification phoneme 5404 5 0 0 5 categorical 2.0 0.1705916225117124 classification -pima 768 8 0 0 8 categorical 2.0 0.0912543402777777 classification poker 1025010 10 0 5 5 categorical 10.0 0.3693735838721754 classification postoperative_patient_data 88 8 2 6 0 categorical 2.0 0.2066115702479339 classification prnn_crabs 200 7 1 0 6 categorical 2.0 0.0 classification -prnn_fglass 205 9 0 0 9 categorical 5.0 0.1061867935752528 classification prnn_synth 250 2 0 0 2 categorical 2.0 0.0 classification profb 672 9 1 2 6 categorical 2.0 0.1111111111111111 classification ring 7400 20 0 0 20 categorical 2.0 9.46676406135857e-05 classification @@ -378,8 +399,7 @@ schizo 340 14 2 0 12 categorical 3.0 0.1478546712802768 classification segmentation 2310 19 0 1 18 categorical 7.0 0.0 classification shuttle 58000 9 0 0 9 categorical 7.0 0.5852340828378914 classification sleep 105908 13 0 1 12 categorical 5.0 0.1484938125639567 classification -solar_flare_1 315 12 7 5 0 categorical 5.0 0.0323255228017132 classification -solar_flare_2 1066 12 4 8 0 categorical 6.0 0.0573341452854562 classification +solar_flare 1066 10 5 5 0 continuous 8.0 0.65729049699214 regression sonar 208 60 0 0 60 categorical 2.0 0.0045303254437869 classification soybean 675 35 1 34 0 categorical 18.0 0.0362678286129266 classification spambase 4601 57 0 0 57 categorical 2.0 0.0449060406200498 classification @@ -408,11 +428,9 @@ titanic 2207 8 1 4 3 categorical 2.0 0.1265128522768823 classification tokyo1 959 44 0 2 42 categorical 2.0 0.0775149209345414 classification twonorm 7400 20 0 0 20 categorical 2.0 6.57414170927761e-07 classification vehicle 846 18 0 0 18 categorical 4.0 0.0004266124179609 classification -vote 435 16 0 16 0 categorical 2.0 0.05179548156956 classification vowel 990 13 2 0 11 categorical 11.0 0.0 classification waveform_21 5000 21 0 0 21 categorical 3.0 8.043999999999987e-05 classification waveform_40 5000 40 0 0 40 categorical 3.0 5.787999999999952e-05 classification -wdbc 569 30 0 0 30 categorical 2.0 0.064939878490615 classification wine_quality_red 1599 11 0 0 11 categorical 6.0 0.2288041188970123 classification wine_quality_white 4898 11 0 0 11 categorical 7.0 0.2119743377393727 classification wine_recognition 178 13 0 0 13 categorical 3.0 0.0125299835879308 classification diff --git a/pmlb/update_dataset_files.py b/pmlb/update_dataset_files.py index 2139aba6f..a9629dc62 100644 --- a/pmlb/update_dataset_files.py +++ b/pmlb/update_dataset_files.py @@ -89,6 +89,7 @@ def datasets_to_update() -> list: updated_datasets = updated_sets['changed_datasets'] updated_metadatas = updated_sets['changed_metadatas'] reviewed_datasets = get_reviewed_datasets(dataset_names) + generate_all_summaries(local_cache_dir=local_dir) for dataset_name in updated_datasets: print(f'Adding readme, metadata and summary stats for {dataset_name}...') diff --git a/tests/test_fetch_nearest_dataset_names.py b/tests/test_fetch_nearest_dataset_names.py index 6c56e5f8e..3a02188ec 100644 --- a/tests/test_fetch_nearest_dataset_names.py +++ b/tests/test_fetch_nearest_dataset_names.py @@ -5,7 +5,7 @@ def test_nearest_dataset_is_itself(): """Tests whether the nearest dataset is itself""" - for test_dataset in ['lupus', 'analcatdata_aids', 'car']: + for test_dataset in ['lupus', 'analcatdata_aids']: df = fetch_data(test_dataset, local_cache_dir='../datasets/') nearest = nearest_datasets(df, task='classification', n=10) print('nearest to',test_dataset,':',nearest)