Regression (Correlation) Analysis
Correlation analysis is a statistical approach to forecasting that seeks to establish a relationship between sales and exogenous variables that affect sales, such as adver- tising, product quality, price, logistics service quality, and/or the economy. Past data on exogenous variables and sales data are analyzed to determine the strength of their relationship (e.g., every time the price goes up, sales of the product go down is a strong negative relationship). If a strong relationship is found, the exogenous variables can then be used to forecast future sales. Corporate, competitive, and eco- nomic variables can be used together in a correlation analysis forecast, thus giving it a broad environmental perspective. Correlation analysis can also provide statisti- cal value estimates of each variable. Thus, variables contributing little to the fore- cast can be dropped.
Correlation analysis is potentially one of the most accurate forecasting tech- niques available, but it requires a large amount of data. These large data demands also make correlation analysis slow to respond to changing conditions. Under- standing the advantages and disadvantages of correlation analysis helps clarify when it is more useful—as in longer-range (greater than six-month time horizon) corporate-level forecasts for which a large amount of data on exogenous variables is readily available.