An explainable machine-learning decision-support system for assessing climate impacts on Kenyan crop productivity using climate, yield, pesticide and NDVI data, with scenario simulation, forecasting and vulnerability ranking.
# Climate–Agriculture Decision Support System
An explainable machine-learning platform for analysing climate impacts on agricultural productivity in Kenya through climate-trend analysis, crop-yield modelling, explainable artificial intelligence, scenario simulation, recursive forecasting, vulnerability ranking, and an interactive decision-support dashboard.
---
## Project purpose
The **Climate Decision Support System** is a reproducible climate–agriculture analytics platform for Kenya. It was developed from the research project **“The Prediction of the Impact of Climate Change on Agricultural Productivity in Kenya Using Machine Learning.”** The project brings together long-term climate records, crop-yield observations, pesticide-use indicators, vegetation information, statistical analysis, machine-learning models, explainable artificial intelligence, climate-scenario simulation, recursive forecasting, and crop-vulnerability ranking in one transparent workflow.
The system is intended to support researchers, agricultural planners, policy analysts, students, and other users who need a national-level view of how climate variability and long-term climate change may relate to agricultural productivity in Kenya. It is a decision-support and research platform rather than a physical crop-growth model, a causal inference system, or a farm-level advisory service.
## System Interface
### National Overview
### Climate Scenario Analysis
### Yield Forecasting
### Vulnerability and Policy Insights
---
## Research Scope
- **Geographical scope:** Kenya
- **Temporal resolution:** Annual
- **Unit of analysis:** National crop-year observations
- **Representative crops:** Sugar cane, potatoes, tomatoes, bananas, and avocados
- **Primary outcome:** Crop yield in hectograms per hectare (`Yield_hg_ha`)
- **Interpretation scale:** National-level patterns; results are not county-level, seasonal, or farm-level predictions
The five representative crops cover different agr …