The evaluation of forecasting techniques as applied to housing starts

Abstract

This thesis is a comparison of different forecasting procedures used in forecasting housing construction. The forecasting models developed are compared on the basis of error variance and the number of times the forecast falls outside "± K" limits. The author points out that what is needed is a simple forecasting technique which gives comparatively accurate forecasts. One such technique is obtained by using the simplest exponential smoothing model, that of single smoothing with a simple adaptive smoothing procedure, as described in this thesis.