Temperature Forecasting for Iwo City, Nigeria Using Statistical Models and Selected Machine Learning Algorithms

Time series modeling and forecasting using a machine learning algorithm approach for Iwo City, a western city in Nigeria, is a technique used to forecast the future occurrence of diverse data for the study. The purpose of this study is to use time series models and some machine learning approaches to predict the temperature in the future for climate change. The Modern-Era Retrospective analysis for Research and Applications, Version 2 (MERRA-2) Web service provided the information used in this study. The dataset underwent training and testing in an 80/20 per cent manner. Using seasonal auto-regressive integrated moving average (SARIMA) and twenty-four (24) different Machine Learning algorithms of regression performance of the temperature, future predictions of Iwo station, were made using training data that spans the years 1980 to 2012 and test data that spans the years 2013 to 2021. The results reveal that the standard deviation value ranges from 2.4×10 -05 to 0.031, while the optimum model for the SARIMA (1, 0, 0) and Akaike’s Information Criterion (AIC) spans from 961.090 to 1341.831. Furthermore, the results of the statistical analysis show that the maximum average temperature recorded in April was 303.68K, the minimum temperature recorded in January was 293.23K with a maximum standard deviation of 1.84 in February, and the minimum standard deviation for the temperature was 0.42 in September. For the machine learning algorithm, Exponential GPR shows the highest R 2 of 0.19 while the least Ensemble Boosted Trees (R 2 = -46.24). In terms of the forecasting performance of these machine learning algorithms based on RMSE, the best forecasting model was Medium Neural Network (RMSE = 1.8112). The result also reveals that Fine Gaussian gave the least MAE = 1.19, while the least MSE = 2.7534 meaning that these machine learning algorithms outperformed other models. As a result, the report recommends that Nigerian meteorological management establishes additional research centers to collect data and conduct research.