gloria-ayesiga/K-means-application

This Python-based tool uses a machine learning technique called K-Means clustering to group soil samples based on their chemical properties. The goal is to identify patterns in soil health and fertility across Sub-Saharan Africa, helping to better understand which areas have similar soil condition # AfSIS Soil Chemistry K-Means Clustering Application A Python-based tool that applies K-Means clustering to soil chemistry data from the Africa Soil Information Service (AfSIS) to identify meaningful soil groups across Sub-Saharan Africa. The application helps reveal patterns in soil fertility, acidity, nutrient availability, and salinity insights useful for agricultural planning, fertilizer recommendations, and sustainable farming practices. ## Features - Loads and preprocesses wet chemistry soil data (pH, Mehlich-3 nutrients, EC, exchangeable bases, etc.) - Performs feature scaling and K-Means clustering (with elbow + silhouette methods to choose optimal k) - Interprets clusters into practical soil profiles (e.g. fertile high-base, acidic low-fertility, saline/high-Na) - Joins georeferenced locations and visualizes clusters on an interactive Folium map - Includes unit tests for core functionality (data prep, scaling, clustering, merging) ## Dataset The project uses the **Africa Soil Information Service (AfSIS) Soil Chemistry** dataset, publicly available on AWS Open Data Registry: `s3://afsis` (no credentials required) Focus: 2009–2013 wet chemistry measurements (e.g. `Wet_Chemistry/CROPNUTS/Wet_Chemistry_CROPNUTS.csv`) Georeferences: `2009-2013/Georeferences/georeferences.csv` License: ODC Open Database License (ODbL) v1.0 – attribution to AfSIS required. ## Requirements - Python 3.8+ - Libraries: pip install pandas numpy scikit-learn matplotlib seaborn folium