The error term is required to satisfy the following assumptions: A negative value of the error term occurs when the actual value of the dependent variable is less than the predicted value. A positive value of the error term occurs if the actual value of the dependent variable exceeds your predicted value ( B 0 + B 1 X 1 + B 2 X 2 + … B n X n). The error term is a random variable that captures the fact that regression models typically do not fit the data perfectly rather they approximate the relationships in the data.
Utilizing multiple regression may lead to improved forecasting accuracy along with a better understanding of the variables that actually cause Y.įor example, a multiple regression model can tell you how a price cut increases sales or how a reduction in advertising decreases sales. Therefore, to gain better and more accurate insights about the often complex relationships between a variable of interest and its predictors, as well as to better forecast, one needs to move towards multiple regression in which more than one independent variable is used to forecast Y. In this chapter the dependent variable Y usually equals the sales of a product during a given time period.ĭue to its simplicity, univariate regression (as discussed in Chapter 9, “Simple Linear Regression and Correlation”) may not explain all or even most of the variance in Y. In causal forecasting, you try and predict a dependent variable (usually called Y) from one or more independent variables (usually referred to as X 1, X 2, …, X n). This chapter continues the discussion of causal forecasting as it pertains to this need. Using Multiple Regression to Forecast SalesĪ common need in marketing analytics is forecasting the sales of a product. Marketing Analytics: Data-Driven Techniques with Microsoft Excel (2014) Part III.