Tutorial and Survey

This paper illustrates the use of the Box‐Jenkins methodology by analyzing the Ohio Electrical Power Consumption time series. The three basic analytical steps discussed are the following: 1) Model identification ‐ by matching sample autocorrelation functions against theoretical autocorrelation functions. 2) Non‐linear estimation of parameters ‐ by minimizing the sum of the squared residuals. 3) Diagnostic checking ‐ by analyzing the pattern of the autocorrelation function of the residuals. Regular and adaptive forecasts are then developed using the appropriate model that emerges from the time series analysis.