Modelling Annual Temperature Using Selected Climatic Variables in Nigeria: An All-Subsets Regression Approach

Abstract

This study examined the relationship between temperature and selected climatic variables, humidity, rainfall, precipitation, and wind using multiple regression modeling to identify the most suitable predictive model for temperature variation. Fifteen possible regression models were estimated and evaluated using regression coefficients, analysis of variance (ANOVA), coefficient of determination (R²), adjusted R², and mean square error (MSE). The results shows that model 10 (X₃X₄), comprising precipitation and wind, recorded a relatively high explanatory power with R² = 0.406 and adjusted R² = 0.368, indicating that approximately 40.6% of the variation in temperature is explained by the model. However, models 12 and 13 demonstrated even stronger performance, with R² values of 0.452 and 0.451 respectively, suggesting better predictive accuracy. Model 13 (Humidity, Precipitation, and Wind) recorded the highest adjusted R² of 0.396 and the lowest MSE of 0.054, indicating that it provides the best balance between explanatory power and prediction error, while humidity and rainfall showed weak individual predictive power. Out of the fifteen estimated models, twelve were statistically significant at the 5% level. Among the models, the combination of humidity, precipitation, and wind produced the best predictive performance with the highest adjusted R² and lowest mean square error, indicating superior explanatory ability. The findings further showed that precipitation had a negative relationship with temperature, while wind had a positive influence. This suggests that temperature variability is largely driven by the interaction of multiple climatic factors rather than isolated variables. The study concludes that multiple regression provides a reliable framework for understanding climatic relationships and can serve as a useful tool for climate prediction and environmental planning. The results provide empirical evidence for improved climate monitoring and policy formulation aimed at mitigating the effects of climate variability.