A Comparative Analysis of Short Term Load Forecasting for Chhattisgarh State

Forecasting addresses the need for demand decisions by using a set of methods that identify future loads in sufficient detail to allow critical decisions to be made. Electricity consumption at any time of the day is called short-term load forecasting. In this paper, an input dataset has been trained, tested with various STLF (Short-term load prediction) methods such as Linear Regression model, boosted tree, Fine tree, Quadratic SVM (support vector machine), Kernel approximation regression model and Artificial Neural Network model (ANN) so as to discover the most accurate model for the prediction of future load. The input dataset mainly consists of generation and load where load demand of former seven days along with the same day generation and load data are arranged in manner so that the load value is estimated for the next day. In input dataset, the data values vary from 25 DEC 2021 to 29 FEB 2024 for overall performance i.e. for both training and testing. Further, evaluation matrices such as MAPE (Mean Absolute Percent Error) and MAE (Mean Absolute Error) are calculated to evaluate the difference between actual and predicted values.