This research evaluated the impact of meteorological conditions on photovoltaic solar panel electricity generation in Kano State, Nigeria, using high-resolution data of 1,051,200 one-minute observations over a two-year timeframe to validate six supervised machine learning models for predictive accuracy: Linear Regression, Ridge Regression, Random Forest, Gradient Boosting, XGBoost, and LightGBM. A feature engineering pipeline was developed with 26 predictor variables, including a clarity index, ambient features, and cyclically encoded temporal variables. The Random Forest model had a high R² of 0.9991, a low RMSE of 9.4791 W/m², and a low MAE of 5.1899 W/m². Cross-validation confirmed model stability (R² = 0.9981 ± 0.0003). SHAP (SHapley Additive exPlanations) analysis identified Global Horizontal Irradiance and module temperature (Module A) as the two most influential predictors. At lower ambient temperatures, Temperature A exhibited an inverse (non-linear) relationship with PV module output; however, at temperatures exceeding 55°C, module efficiency declined due to the negative temperature coefficient of crystalline silicon. Upon comparison of the mean Global Horizontal Irradiance for Kano throughout the wet (June through September) and dry months (December through February); it is seen that there was a reduction of approximately 36% in average GHI from 450 – 500 W/m2 to 700 – 750 W/m2 at noon. These findings offer critical insights for regional energy planning, infrastructure resilience, and the optimization of solar integration in Northern Nigeria’s grid.