A forecasting model addressing the issue of correlation with application to automobile sales

Includes bibliographical references (pages 63-64)Forecasting is a method of predicting future events\ud from historical outcomes. Forecasts are not free from\ud error. However, error can be minimized through the careful\ud analysis of data and selection of relevant variables\ud for inclusion in forecasting models. Multicollinearity\ud has presented problems with respect to correlation among\ud the independent variables. Many times the analyst assumes\ud independence among the predictor variables and allows\ud errors due to multicollinear relationships to be \ud accumulated with other sources of random error. Factor\ud analysis is a statistical technique used in the social\ud sciences to examine correlative relationships in variables.\ud The outcome of this analysis procedure is a minimum\ud set of independent factors that explain the multicollinear\ud relationships in the variables. By utilizing\ud these factors in a regression model a less costly forecast\ud can be developed which accounts for correlation\ud among the variables in terms of the independent factors.\ud The decision for usage of a forecast is primarily\ud based on the benefits derived in increased profits versus\ud the cost of developing a more realistically modeled forecast.\ud It is assessed in this thesis that overstated\ud forecasts versus understated forecasts are deemed more\ud favorable in cases where inventory holding costs are lower\ud that stock-out costs. The reverse is true when stock-out\ud costs are lower than inventory costs. \ud A case study was presented for analysis regarding\ud the issue of automobile sales. The problem identified\ud from review of market activity was an unrecognized change\ud in demand for full-size cars relevant to the auto industry.\ud Before implementing a change in production to increase\ud sub-compact units, the characteristics and amount\ud of demand for sub-compact cars should be forecast.\ud By developing an accurate forecast of demand for subcompact\ud vehicles, American automobile manufacturers will be \ud able to minimize additional costs incurred by changed\ud production processes for the modified product mixture\ud (increase production of sub-compacts, decrease production\ud of full-sized cars).\ud An analysis was conducted on data available from the\ud Department of Commerce. Results of the analysis indicate\ud that a factor analysis would be appropriate for development\ud of the forecasting model for sub-compact car demand.\ud The risk of increasing holding costs due to an overstated\ud forecast were considered worthwhile in consideration of\ud the profits that would be lost by adhering to an understated\ud forecast.\ud The example used in this thesis demonstrates the use\ud of factor analysis and supports the implementation of\ud this technique for model development of sub-compact car\ud demand. In consideration of the successful utilization\ud of this procedure, recommendations were made to pursue\ud research toward the practical application of factor\ud analysis in forecasting methods