jaysplash208/Oil-palm-yield-weather-analysis

This project applies time series analysis to 30 years of meteorological data from NIFOR stations in Nigeria to forecast maximum and minimum temperatures. The goal is to provide insights that can support oil palm yield planning and agricultural decision-making. # Oil-palm-yield-weather-analysis Project Overview This project focuses on the Forecasting of maximum and minimum temperatures and other meteorological data at the NIFOR Main Station and various substations across Nigeria. By analyzing 30 years of historical time-series data, this model provides critical insights into climate patterns that directly influence oil palm yield. The goal was to build a robust predictive system that assists in agricultural planning and sustainability for the Nigerian Institute for Oil Palm Research. # The "Model Showdown" I implemented a hybrid approach to find the most accurate forecasting method: 1. SARIMA (Seasonal Auto-Regressive Integrated Moving Average): A classical statistical model used to capture seasonal climate trends. 2. Random Forest Regressor: A modern Machine Learning approach used to handle non-linear relationships in weather patterns. # The Result * Winner: Random Forest Regressor * Final Accuracy: Mean Absolute Error (MAE) of 1.10°C * Performance: The ML approach significantly outperformed the classical SARIMA model (MAE 4.13°C) for this specific 30-year dataset. # TECH * Language: Python * Environment: Google Colab * Libraries: pandas, numpy, scikit-learn, statsmodels, matplotlib, joblib # Key Features * Substation Analysis: Designed to be adaptable for different geographical substations. * Long-term Forecasting: Capable of projecting temperature trends based on three decades of data.