This is the source code of the method published in "Enhancement of Consumption Forecasting by Customers’ Behavioral Predictability Segregation". In our research we developed the method of customers separation by predictability and its dynamic with the help of LSTM models. Our research shows that (1) customer separation helps to improve forecasting quality of the whole population due to the decomposition of all clients’ time series, (2) environment instability such as pandemic or military action can be leveled out with incremental models.

File Summary