Retail companies need to predict the demand for their products in each shop for every day of the year. Usually, the sales planning is done by taking basic statistical operations of the last months of sales that don’t always catch the trend and the seasonality of the purchases to predict the future demand.
We developed a demand forecasting model for a retail client based on historical data and outside variables using advanced time series models & analysis.
The advanced state-of-the-art time series model was implemented to predict the demand on 14.000 different products in 150 different stores by combining three different modeling techniques, LSTM, Gradient Boosting Models, and ARIMA.
Stand-alone solution for demand forecasting of each product in each store combined with the integration of the current stock for better order planning.