FORECASTING THE INVESTMENT INDICATORS ON THE BASIS OF SOLUTION TREES

Abstract

The article shows specific features of short-term forecast models, which were realized on the basis of the solution trees method applied to regression task. The authors provide the description of key characteristics of the trees solution method using CART algorithm. The necessity to use alternative intellectual methods of forecasting was grounded due to more difficult and highly uncertain nature of political and social-economic processes. The mechanism of final summing-up of forecasts was described. The authors put forward several variants of implementing models designed by Python programming. Computer experiments were conducted to adjust model parameters that can guarantee acceptable results in forecasting investment indicators.

Keywords

• investment indicators
• short-term forecasting
• cart algorithm
• system of hybrid models