The Use of Statistics in Business Forecasting
By LEWIS H. HANEY and FREDERICK C. REICH
This chapter attempts the difficult and perhaps impossible feat of presenting in brief form a practical discussion of the statistical methods most useful in business forecasting, without discussing the general elements of statistics. It will be dry and difficult reading for one who is not interested in the practical work of the forecaster. It will, in part, seem unduly simple to the experienced statistician. The practical forecaster, however, will find of real interest several points based on the authors' experience. The student, under the guidance of his teacher (who may well suggest exercises to illustrate the main points touched upon), will find it essential to an understanding of business forecasting to gain an accurate conception of "relatives," "seasonal variation," "long-time trend," "correlation," and other expressions. This is the reason for presenting such a long, and at points difficult, chapter so early in the book. For the most part, numerical illustrations and technical demonstrations have been relegated to an Appendix, which should be consulted by those interested.
Forecasting deals chiefly with "time series," that is, statistics showing trends from month to month or other periods of time. Upon methods used in handling these series largely depends the success of the forecaster.
Few statistical series are considered by themselves; rather one series is compared with others, usually interrelated, in order that significant relationships may be brought out, either as sequences or as maladjustments. To attain this objective