The aim of this study is to estimate Solar Radiation (SR) values for various data packages in Turkey using satellite data for 3600 grid points and 81 provinces points processed with the HELIOSAT model covering the period 2004–2021 obtained from the General Directorate of Meteorology (MGM). In this study, the Model 5 Tree (M5-Tree) heuristic regression method was used to predict solar radiation values. Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Relative Error (MARE), Nash–Sutcliffe Equation (NSE), and coefficient of determination (R2) were used as evaluation criteria. The Inverse Distance Weighting (IDW) interpolation method was used for SR maps. In order to investigate the effect of the obtained prediction model, the obtained results were evaluated by graphical comparison (time series graphs and scattering graphs). The observation data to be used in the M5-Tree method was divided into 4 separate packages, 75% training, and 25% testing. In the test phase, which was analyzed as 2 inputs (latitude-longitude) and 1 output (SR value) on the MATLAB platform, the best and worst results were RMSE: 0.0850, MAE: 0.0636, MARE: 1.4312, NSE: 0.9637 and R2: 0.9643 and RMSE: 0.0995, MAE: 0.0691, MARE: 1.5636, NSE: 0.9502 and R2: 0.9506. When the evaluation criteria were analyzed, it was seen that all data packages had values close to each other. All data packages were successful in 4 different packages with the prediction values obtained. In addition, it is thought that it will be even more successful with new parameters added to the input parameters used. As a result, the M5-Tree method for SR estimation gave very successful results and the model using more data (3600 grid points) is more successful.