Climate prediction models are crucial for understanding climate patterns and their impacts on various sectors in Kenya. Machine learning (ML) techniques have shown promise in enhancing these predictions by leveraging complex data sets, but there is a need to validate and standardise methods for practical application. The study employs a comparative analysis of several ML models, including Random Forest, Support Vector Machines (SVM), and Neural Networks, trained on datasets from multiple climate stations across Kenya. Model selection is guided by cross-validation techniques to ensure robustness and generalizability. Performance metrics such as Mean Absolute Error (MAE) are used for model evaluation. Random Forest models demonstrated the lowest MAE of 3. 5°C in temperature predictions, indicating high accuracy with a confidence level of within ±1. 2°C, suggesting their suitability for climate adaptation planning. This study validates the efficacy of ML models in improving climate prediction in Kenya and provides a framework that can be adapted to other regions facing similar challenges. Future research should focus on integrating real-time data sources and expanding model validation across diverse geographical areas within Kenya. Machine Learning, Climate Prediction, Adaptation Planning, Random Forest, Support Vector Machines Model estimation used =argmin_ᵢ (yᵢ, f_ (xᵢ) ) +₂², with performance evaluated using out-of-sample error.
Mbengo et al. (Sun,) studied this question.