Multisteps Sales Forecasting using LSTM Compared to GMDH
Ghita Rguiga, Jamal Benhra · International Journal of Computer Applications · 2021
This work reports the effectiveness of using Recurrent Neural Network LSTMs for multistep sales forecasting instead of a standalone approach and the use of time series data, treatment of inflation, forecasting options and tunings to help decision executives make better decisions on forecast and management systems.This work is lead on historical sales based on a year of demand records, the program established using LSTMs showed minimal loss and optimal forecasts compared to prediction using GMDH which is known for management forecasts such as inventories and sales forecasting.