Analysis of Telephone Data: A Case Study of Forecasting Seasonal Time Series

In this paper, forecasting models are constructed for the monthly inward and outward station movements of the Wisconsin Telephone Company using an iterative procedure developed by Box and Jenkins. Data covering the period January 1951 through October 1966 were used to develop the model. Forecasts with 95-percent probability limits were calculated for three years from November 1966, and were compared with the actual observations. The properties of the models are discussed in detail. Alternative models for forecasting are also entertained and compared with those chosen.