Nepal District-Level Maize and Wheat Yield and Climate Panel Dataset (1981–2023)

This dataset accompanies the research article 'Developing a Machine Learning-Based Early Warning System for Maize and Wheat Yield Forecasting in Nepal' submitted to the International Journal of Agricultural Sustainability (Taylor & Francis). It comprises a balanced panel of 77 Nepalese districts over 43 years (1981–2023), containing 3,311 district-year observations per crop with 30 variables spanning: district-level maize and wheat yields (kg/ha) from MoALD SINA reports; multi-source climate variables including rainfall, temperature, dry days, and extreme precipitation events from DHM Nepal and ERA5-Land reanalysis; standardized climate anomalies and Standardized Precipitation Index (SPI-3) scores for both kharif and rabi seasons; MODIS-derived NDVI composites (2001–2023, MOD13A3 v6.1); and engineered predictive features including lagged yields, monsoon onset lag, and disaster and yield shock binary flags. The dataset was used to train and benchmark three machine learning models (Random Forest, Support Vector Machine, Decision Tree) with SHAP-based feature attribution analysis, and to calibrate a four-tier agricultural early warning system for Nepal's food security governance.