On modeling time series data using spreadsheets

Linear regression analysis has long been used to estimate the parameters of various types of times series (TS) models. In some cases, the application of traditional regression-based techniques to TS data does not produce the optimal values of the parameters being estimated. The nonlinear optimizatio n tool (known as Solver) built into today's electronic spreadsheets can all eviate these estimation problems as well as simplify the process of modelin g many types of TS problems. This paper presents two examples of TS problem s where Solver performs better than regression-based TS techniques. It also encourages educators to reevaluate these and other traditional TS techniqu es in light of Solver's capabilities. (C) 2000 Elsevier Science Ltd. AII ri ghts reserved.