10 History Cleansing Best Practices for Reliable Statistical Forecasting
- by Rajesh Ray, Senior Managing Consultant and SCM Product Lead, IBM Global Business Services
- December 9, 2011
Generate the most reliable and accurate sales data history for your statistical forecasting as part of your SAP Advanced Planning & Optimization Demand Planning project. See how to implement an efficient and effective history cleansing process.
The baseline history of a product is its normal historical demand without promotion, external stimulation, or any other abnormal situation. An outlier is a too-high or too-low sales figure in a product’s history that may occur under special conditions.
History cleansing is the process of cleansing the history of sales data, which is a prerequisite when a company is using the Statistical Forecasting functionality of SAP Advanced Planning & Optimization (SAP APO) Demand Planning (DP). The better the history is, the better the forecast results.
The purpose of history cleansing is to produce a baseline history (i.e., the sales data for a product during its normal lifespan when there are no promotions, shortages, or any other unexpected or abrupt market conditions or changes). Statistical forecasting tools attempt to identify trends that might repeat in the future. Because events such as shortages or product promotions cannot always be predicted, that related data needs to be eliminated from the history used in forecasting.
Though promotions are often regular occurrences for consumer goods companies, these events might not always happen in the same time frames (i.e., in the same weeks or month in the future) as they did in the past, or the durations for such promotions might also be different. Once the sales history is cleansed, forecasting uses a baseline history, called baseline forecast. After baseline forecast numbers have been obtained, you can add data from future promotions and other events to the baseline forecast to get final forecast numbers.
Would you like to see this full item?