Global warming is becoming a challenging issue in the world that requires adequate and appropriate attention more especially in arid and semiarid regions. The aim of this study was to assess the potential of data mining algorithms in forecasting the climate change impact on temperature for Sokoto state, Nigeria. In this way, Hammerstein Wiener (HW), Nonlinear Autoregressive Exogenous (NARX) and Autoregressive Integrated Moving Average (ARIMA) models were employed for statistical downscaling and future mean monthly temperature projection for Sokoto station, Nigeria. For this purpose, global circulation models (GCMs) from Coupled Model Intercomparison Project Phase 5 (CMIP5) were used as the downscaling predictors while the predictand were collected from 1990–2022. Since the selection of the most dominant predictors is of paramount importance for any artificial intelligence-based prediction, the correlation coefficient (CC) method was used to determine the appropriate predictors. Based on CC results, 5 different models were developed to ensure high downscaling efficiency. The mean monthly temperature projection was then performed from 2067–2099 according to the best downscaling model under the RCP4.5 scenario. The future projection results showed that Sokoto station would experience both an increase and decrease in temperature. An increase of up to 3.5% would be experienced between December – January and a maximum decrease of 2% would be expected between March – April. This study outcome could be useful in water resource management, as well as decision-making and climate change mitigation.