Maryam-Shile/weda.io

A climate prediction and crop advisory app **Weda.io-A Climate Prediction & Crop Advisory App** **Overview** Weda.io is a climate prediction and crop advisory application built with African farmers and agricultural investors in mind. Extreme weather events continue to cause severe agricultural losses across Africa. For example, the Food and Agriculture Organization (FAO) documented approximately ₦4 trillion worth of food loss in Benue State in 2024 due to flooding in major agricultural hubs. Climate variability has increasingly made agricultural investment unpredictable. Weda.io aims to reduce that uncertainty. This pilot project uses historical climate data to: - Predict total precipitation over a selected period - Predict average temperature over a selected period - Assess crop–climate compatibility-Provide actionable advisory insights for crop planning **Problem Statement** Agricultural investment in many African regions is highly vulnerable to: - Flooding - Irregular rainfall patterns - Temperature shifts - Climate change–driven variability Farmers and investors often operate without localized predictive tools tailored to their specific agro-ecological conditions. **What Weda.io Does** This pilot implementation is built using Climate Data Store (CDS) data collected in: Mokwa, Niger State, Nigeria Climate Prediction Rainfall Prediction - Model: SARIMA - Mean Absolute Error (MAE): 24.13 Temperature Prediction - Model: AutoETS - Mean Absolute Error (MAE): 0.498 Users select a forecast period, and the system generates predicted: - Total precipitation - Average temperature - Crop Compatibility Engine After generating predictions, the system: - Evaluates the climate conditions - Compares them to crop suitability thresholds - Produces a structured advisory message on crop viability **Demo** You can explore the project here: weda-predict.streamlit.app **Model Details** **Rainfall Model** Type: Seasonal ARIMA (SARIMA) Seasonal length: 12 Performance metric: MAE = 24.13 Currently under optimizat …