Sales Forecasting
As the brand manager for a packaged goods company, you want to know how many cases of shampoo will be sold in grocery stores in the Pacific Northwest next month. You also want to know whether the sales of styling gel affect shampoo sales – and vice versa. And, you want the forecast to take seasonality into account.
Data mining answers questions like these much better than traditional time series forecasts, which don’t allow for other variables such as other product sales, seasonality, or past promotions.
E.J. Barry uses a new time series algorithm developed by Microsoft® Research, called autoregression trees, or ART. The benefits of ART for sales forecasting include:
- Forecasts for multiple products in the same model
- Determining how sales of one product may affect sales of another in the same forecast period
- Testing of forecast predictions on known historical sales data (how well the model predicted sales results in the past)
- Automatic detection of seasonality effect on sales
Accurate forecasting is increasingly important to avoid stock outages and unused inventory. Our data warehouses are the perfect platform for sales forecasting because they include time dimensions and historical data about sales, customers, and products.
Contact us to find out how we can help you improve the accuracy of your forecasts.
