Developed demand forecasting AI for home appliance manufacturer (West Coast)

Background

Our client, a manufacturer that sold consumer products in the United States, made monthly sales forecasts for hundreds of items. Since products manufactured at a Japanese factory were temporarily stocked before being sold, if there is a large difference between sales forecasts and actual results, cost increases or opportunity losses would occur.

Challenge/Issue

Based on the sales volume of last year and sales forecasts from sales interviews, the client created Excel files of the predicted sales. The client tried to reduce the prediction error and workload necessary for the calculation. 

Solution

Based on the sales data for the last 10 years or more, we developed a prediction model for each item and an application for the prediction operation. This application enables the person in charge to create the prediction file automatically just by inputting the sales data and the prediction period. 

Result

Compared to the previous process, the prediction error was reduced to 1% for a certain item. Plus, the application reduced the necessary time for the prediction from 1month to several hours.