• Story Name: Predicting Retail Sales
  • Story Topics: Economics, Consumer,
  • Datafile Name: Predicting Retail Sales
  • Methods: Regression, Residuals, Time Series,
  • Abstract: The datafile contains 11 years of quarterly sales for four kinds of retail establish- ments, along with non-agricultural employment and wage and salary disbursements The task is to develop a model for predicting sales using leading values of employ- ment or wage and salary disbursements, seasonal indicators or seasonal lags, other lags of the dependent variable, and time. If significant positive autocorrelation is present in the residuals from the best model, a revised model can be obtained by introducing lagged values of the residuals into the model as an additional predictor. Use a prediction routine to obtain predictions for the four quarters of 1990 from the best model, and the 1st quarter of 1990 for the model if modified by use of lagged residuals. Compare the prediction with the actual value(s) for 1990 given below. Note whether the 95% confidence intervals for the prediction(s) include the actual value(s). Values for 1990:
               Quarter          BDLG        AUTO        FURN        GMER
                    1              19368         92253       21738        41832
                    2              26220       103038       22842        50181
                    3              99006         22620       49137        34911
                    4              87063         25611       71265        36543 

Contact Us

© 2017 Data Description, inc. All rights reserved.