Djiby223/Mali_Crop_Yield_Prediction

This project builds a machine learning model to predict cereal crop yield in Mali using real agricultural statistics from FAOSTAT and climate indicators such as rainfall and drought conditions. The goal is to demonstrate how climate variability impacts agricultural productivity and to build a reproducible AI pipeline for agricultural forecasting. # Mali_Crop_Yield_Prediction This project builds a machine learning model to predict cereal crop yield in Mali using real agricultural statistics from FAOSTAT and climate indicators such as rainfall and drought conditions. The goal is to demonstrate how climate variability impacts agricultural productivity and to build a reproducible AI pipeline for agricultural forecasting. Objectives • Clean and process FAOSTAT agricultural yield data • Integrate climate data (rainfall + SPI) • Build a supervised machine learning model • Predict crop yield using climate indicators • Evaluate model performance using regression metrics ________________________________________ 📊 Datasets Used 1. FAOSTAT Agricultural Data Source: FAOSTAT (Food and Agriculture Organization) • Area: Mali • Item: Cereals, primary • Variable used: Yield • Time range: ~1961–2024 📄 File: FAOSTAT_data_en_6-28-2026.csv ________________________________________ 2. Climate Data (SPI + Rainfall) • Annual rainfall (mm) • Standardized Precipitation Index (SPI) 📄 File: mali_spi_results(2).csv ________________________________________ 🧹 Data Processing Workflow Step 1: FAOSTAT Cleaning • Removed irrelevant columns • Filtered Element = Yield • Renamed target column to yield_t_ha • Kept only cereals data Step 2: Climate Aggregation • Converted monthly SPI to yearly values • Aggregated rainfall and SPI by year: o Annual rainfall mean o Annual SPI mean Step 3: Data Merge Merged datasets on: Year Final dataset: mali_crop_yield_climate_merged.csv ________________________________________ 📦 Final Dataset Structure Year yield_t_ha annual_rainfall_mm annual_spi 1981 837.9 90.48 -0.146 1982 785.0 86.00 0.035 ________________________________________ 🤖 Machine Learning Model Model Used • Random Forest Regressor (scikit-learn) Features (X) • annual_rainfall_mm • annual_spi Target (y) • yield_t_ha ________________________________________ ⚙️ Model Pipeline 1. Train/test split (80/20) 2. Model training 3. Prediction 4. Evaluation ___ …