coorddetect

coorddetect automatically detects X, Y, Z coordinate columns from messy tabular data (CSV and Excel) using a robust, file-order–preserving heuristic. It also provides optional spatial QA/QC diagnostics such as robust bounding boxes, convex hull geometry, and point density metrics.

The XYZ detection logic does not rely on column names and is designed for real-world geomatics applications, GIS, and survey datasets.

---

Installation

pip install coorddetect

---

Quick Usage

import pandas as pd
from coorddetect import detect_xyz

df = pd.read_csv("points.csv")

xyz, meta = detect_xyz(df)

print(xyz.head())
print("Selected XYZ columns:", meta["selected_columns"])

---

Preserve an ID Column

xyz, meta = detect_xyz(
    df,
    id_col="WallStationID"
)

Output columns

ID, X, Y, Z

---

Spatial Diagnostics (All Features)

By default, detect_xyz() can return:

• Robust bounding box
• Convex hull geometry
• Point density metrics

xyz, meta = detect_xyz(
    df,
    return_bounds=True,
    robust_bounds=True,
    bounds_quantiles=(0.01, 0.99),
    return_hull=True,
    hull_dim=2,            # 2 = XY hull, 3 = XYZ hull
    return_density=True
)

print("Bounds:", meta["bounds"])
print("Hull dimension:", meta["convex_hull_dim"])
print("Density:", meta["density"])

---

Robust vs Raw Bounding Box

# Outlier-resistant (recommended)
xyz, meta = detect_xyz(
    df,
    robust_bounds=True,
    bounds_quantiles=(0.01, 0.99)
)

# Raw min / max bounds
xyz, meta = detect_xyz(
    df,
    robust_bounds=False
)

---

Disable Diagnostics (Pure XYZ Detection)

xyz, meta = detect_xyz(
    df,
    return_bounds=False,
    return_hull=False,
    return_density=False
)

---

Batch Processing (CSV & Excel)

Process all .csv, .xlsx, and .xls files in a folder.

from coorddetect import detect_xyz_batch

results = detect_xyz_batch(
    input_folder="data/raw",
    output_folder="data/out",
    recursive=True,
    export_format="csv",     # or "xlsx"
    excel_sheet=None,        # sheet name or index for Excel
    id_col="WallStationID",
    return_bounds=True,
    robust_bounds=True,
    return_hull=True,
    hull_dim=2,
    return_density=True
)

print(results)

Each file is exported as:

<original_name>_xyz.csv

---

Command Line Interface (CLI)

After installation, the CLI is available:

coorddetect INPUT_FOLDER OUTPUT_FOLDER

Common Examples

# Process a folder recursively
coorddetect data/raw data/out --recursive

# Export Excel instead of CSV
coorddetect data/raw data/out --export-format xlsx

# Use raw min/max bounds
coorddetect data/raw data/out --raw-bounds

# Use 3D convex hull
coorddetect data/raw data/out --hull-dim 3

# Disable hull and density metrics
coorddetect data/raw data/out --no-hull --no-density

View all options

coorddetect --help

---

Notes

• Convex hull and point density metrics require scipy
• Convex hull requires at least 3 points (2D) or 4 non-coplanar points (3D)
• All diagnostics are computed after XYZ detection
• XYZ detection logic does not rely on column names

---

Applications

• GIS applications
• Survey data validation
• Geomatics research pipelines

---

License

MIT