Commit Live PR

q01_plot_corr/build.py

@@ -1,5 +1,7 @@
+# %load q01_plot_corr/build.py
 # Default imports
 import pandas as pd
+import matplotlib.pyplot as plt
 from matplotlib.pyplot import yticks, xticks, subplots, set_cmap
 plt.switch_backend('agg')
 data = pd.read_csv('data/house_prices_multivariate.csv')
@@ -9,8 +11,11 @@
 def plot_corr(data, size=11):
     corr = data.corr()
     fig, ax = subplots(figsize=(size, size))
-    set_cmap("YlOrRd")
+    set_cmap('YlOrRd')
     ax.matshow(corr)
     xticks(range(len(corr.columns)), corr.columns, rotation=90)
     yticks(range(len(corr.columns)), corr.columns)
     return ax
+
+
+

q02_best_k_features/build.py

@@ -1,12 +1,30 @@
+# %load q02_best_k_features/build.py
 # Default imports
 
 import pandas as pd
+import numpy as np
+from sklearn.feature_selection import SelectPercentile,f_regression
 
 data = pd.read_csv('data/house_prices_multivariate.csv')
 
-from sklearn.feature_selection import SelectPercentile
-from sklearn.feature_selection import f_regression
+# Write your solution here:
+
+def percentile_k_features(data, k = 20):
+    X = data.drop('SalePrice',axis=1)
+    y = data['SalePrice']
+
+    feat_col = X.columns
+    fs = SelectPercentile(f_regression, percentile=k)
+
+    X_new = fs.fit_transform(X, y)
+
+    imp_features_kth_percentile = [feat_col[i] for i in np.argsort(fs.scores_)[::-1]]
+
+    #print (imp_features_kth_percentile[:7])
+
+    return imp_features_kth_percentile[:7]
+
+percentile_k_features(data,20)
 
 
-# Write your solution here:
 

q03_rf_rfe/build.py

@@ -1,3 +1,4 @@
+# %load q03_rf_rfe/build.py
 # Default imports
 import pandas as pd
 
@@ -6,6 +7,21 @@
 from sklearn.feature_selection import RFE
 from sklearn.ensemble import RandomForestClassifier
 
-
 # Your solution code here
 
+def rf_rfe(data):
+    X = data.drop('SalePrice',axis=1)
+    y = data['SalePrice']
+    random_forest_model = RandomForestClassifier()
+
+    rfe = RFE(random_forest_model,n_features_to_select=len(X.columns)/2)
+    rfe = rfe.fit(X,y)
+
+    #print (list(X.columns[rfe.support_]))
+
+    return list(X.columns[rfe.support_])
+
+rf_rfe(data)
+
+
+

q04_select_from_model/build.py

@@ -1,3 +1,4 @@
+# %load q04_select_from_model/build.py
 # Default imports
 from sklearn.feature_selection import SelectFromModel
 from sklearn.ensemble import RandomForestClassifier
@@ -6,5 +7,21 @@
 
 data = pd.read_csv('data/house_prices_multivariate.csv')
 
-
 # Your solution code here
+
+def select_from_model(data):
+    X = data.drop('SalePrice',axis=1)
+    y = data['SalePrice']
+    rf_model = RandomForestClassifier()
+
+    select_fm = SelectFromModel(rf_model)
+    select_fm.fit_transform(X,y)
+
+    #print (list(X.columns[select_fm.get_support()]))
+
+    return list(X.columns[select_fm.get_support()])
+
+select_from_model(data)
+
+
+

q05_forward_selected/build.py

@@ -1,10 +1,46 @@
+# %load q05_forward_selected/build.py
 # Default imports
 import pandas as pd
 from sklearn.linear_model import LinearRegression
-
+from sklearn.metrics import r2_score
+import numpy as np
 data = pd.read_csv('data/house_prices_multivariate.csv')
 
 model = LinearRegression()
 
-
 # Your solution code here
+
+def forward_selected(data,model):
+    old_r2_score = 0
+    new_r2_score = 1
+    features = list(data.drop('SalePrice',axis=1).columns)
+    selected_features = []
+    r2_score_features = []
+    X_selected = pd.DataFrame()
+    result = pd.DataFrame()
+    y = data['SalePrice']
+    while(True):
+        scores = []
+        for i in range(len(features)):
+            X = data[features[i]]
+            X_selected = result
+            X_selected = pd.concat([X_selected,X], axis=1)
+            model.fit(X_selected,y)
+            y_pred = model.predict(X_selected)
+            scores.append(r2_score(y,y_pred))
+            X_selected = result
+            np_scores = np.array(scores)
+        new_r2_score = np_scores.max()
+        if(new_r2_score>old_r2_score):
+            old_r2_score=new_r2_score
+            result = pd.concat([result,data[features[np.argmax(np_scores)]]], axis=1)
+            data = data.drop(features[np.argmax(np_scores)],axis = 1)
+            selected_features.append(features[np.argmax(np_scores)])
+            r2_score_features.append(new_r2_score)
+            features.remove(features[np.argmax(np_scores)])
+        else:
+            break
+    return selected_features,r2_score_features
+
+
+