Electric power has become a primary human need today, making it essential to maintain the reliability of the system. Therefore, a well-prepared and thorough Masterplan for the distribution system is necessary. The purpose of this research is to analyze the characteristics of the electrical profile using classification results and to develop a load forecasting model for each regional cluster. One of the step in preparing such a comprehensive plan is to perform electric load forecasting modeling. In developing this model, several stages are carried out beforehand. The first stage involves classifying regions based on their electrical characteristics. The classification is performed using the Support Vector Machine (SVM) method, utilizing data that has been previously clustered. Once all the regional groups are obtained, the model can be developed through a regression process. To determine the best model representing each cluster, two different methods are used: Multiple Linear Regression and Random Forest Regression The classification accuracy achieved is 0.83, allowing the model to be applied to new data. The best models for clusters 1 and 4 were obtained using Random Forest Regression, with R² values of 0.9458 and 0.8317, respectively, while for clusters 2 and 3, Multiple Linear Regression was used, with R² values of 0.9185 and 0.9808, respectively. With the establishment of the best model for each cluster, it is expected that the process of designing the power system will be simplified, enabling the development of a reliable and well-structured Masterplan.