Electrical Load Forecasting

Since the electricity demand is increasing globally, load forecasting techniques have become immensely important in forecasting the electricity demands and it also helps the policy makers. The aim of our project is to perform short-term load forecasting, i.e. up to a week ahead. Household owners can estimate the upcoming load and power distribution organization would know the demand and could be prepared henceforth. Our attempt is to generate useful insights and forecast as accurately as possible. We are using different techniques starting from Naive Bayes, Classical Linear Methods (like ARIMA), and some Machine Learning Algorithms (like LinearRegression, Ridge, Lasso, HuberRegressor, ElasticNet, Lars, LassoLars, PassiveAggressiveRegressor, RANSAC Regressor, SGD Regressor) to make predictions. And we are also using Deep Learning algorithms like CNN, LSTM and combining CNN-LSTM to get more accurate predictions. In the end we will compare all the predictions from all the models that we have used and determine which model makes the best prediction.