A Computer program based on the genetic approach has been developed by the author as a supplementary tool for business cycle research. The program applied to a set of cross-section or time series data tries to discover the most probable relationships between economic variables and to determine their impact on cyclical developments in the economy. Unlike in standard econometric models, the point here is not merely to estimate parameters of certain specified equations, but also to identify the proper structure of the model. Preliminary tests of the program, on a set of variables entering the composite leading indicator for Poland deve!oped by Z. Matkowski, indicate a significant analytical potential of the program.
An attempt was made to use the decision tree generator and classifier C4.5 developed by J.R. Quinlan to identify typical sequences between various economie variables in the Polish economy. The C4.5 algorithm was used to discover typical sequences in the cyclical change of some component variables of the composite leading indicator for Poland developed by Z. Matkowski and his reference index (generał coincident indicator - GCI). The research was made on full time series of monthly data and on the shortened time series, cut down to cover the same period between January 1987 and December 1997. Empirical pattems of cyclical change established in the study are compatible with the economic significance of individual variables and the expected relationships among them. Some of these rules can be used in short term forecasts of generał business activity.
JavaScript is turned off in your web browser. Turn it on to take full advantage of this site, then refresh the page.