WILMORE PAPER COMPANY

Advertising Budget Model

Time Series Analysis by Burcu Arikan, Spring 2008

Validation

The final step in the model building process is to assess generalizability of each model in predicting future sales. An "Acid Test" is conducted for each of the nine models to calculate the average percentage error per model. The predictive model is built using sales data from Fall 2000 to Winter 2008, while data for the 3 quarters beginning Summer 2007 to Winter 2008 are reserved. Actual advertising expenditures from Summer 2007 to Winter 2008 are entered into each model to test their predictive validity. Average Percentage Errors (APE) are calculated by comparing the actual and predicted sales.

Average percentage error is calculated for each predictive model to test validity using the equation APE = Actual Sales - Predicted Sales/Actual Sales. Based on the verification and validation phases taken comparatively, the Logistic function is validated for best fit the Wilmore Paper Company data for predicting future sales from advertising expenditure.

Refer to the table in the Estimation section of this website for model comparison results.

Selected Model: Logistic Function

Predicted Sales =

The Scatterplot below illustrates the predicted sales from the Logistic Function.