The aim of this study is to estimate the future electricity demand for domestic and commercial purpose. With the rising demand for power at households and industrial levels, it is more critical and important than ever to estimate future electricity needs so that demands in future can be met. In this paper, the ARIMA forecasting model with machine leaning techniques is presented for electricity demands forecasting. Time series decomposition is used to understand and split the data into test and train. ARIMA model is also compared to some similar models and benefits of using ARIMA model are also discussed. The results of this study show that ARIMA model can be used for forecasting electricity demand with lesser train and test error values as 0.10 and 0.04 respectively.