Sector-wise expert input selection for electricity load forecasting

In Smart Grid, forecasting power demand plays a vital role in demand side management. Forecasting medium-term load is important for power system planning. People depend on electricity for livelihood in the residential sector and for good productivity in the commercial, industrial and agricultural sectors. The sector-wise importance of characteristics for accurate load forecasting has been identified in this paper. First, we analyze the power consumption data in various sectors with different parameters, then various parameters (weather data and seasonal data) are identified that affect the power consumption pattern of the region, finally, a machine learning model is used to identify the sector-wise feature importance for load prediction. The Random Forest Regressor outperforms Multiple Linear Regression for load prediction.