|
Original research SOLVING THE PROBLEM OF MATHEMATICAL MODELING OF THE INFLUENCE OF FACTORS ON THE CONVERSION RATE OF FUNNELS TRANSFER OF INNOVATIVE DEVELOPMENTS USING MACHINE LEARNING TOOLSPages 15-18 Abstract
The article is aimed at analyzing the possibilities of solving the problem of modelling the influence of various factors on the indicator of conversion of the transfer funnel of innovative developments. In the authors' previous publications this indicator was marked as «conversion rate» (percentage of developments that have passed from the stage of «object of intellectual property, received legal protection» to the stage of «patented result of intellectual activity, put on the market»), the list of factors that influence it is determined. This article presents the result of testing the algorithm of modeling with the use of machine learning tools. The information base of the study became open information about indicators of innovation activity and patent activity in the Russian Federation. Research was asked about the possibility of using machine learning tools to construct regression problems describing the influence of factors on a feature, and to draw conclusions about the most important factors. In the course of solving the problem, the following stages of data analysis for machine learning were implemented: cleaning and formatting, preliminary analysis, selection of the most important factors, model testing on a test sample. In conclusion, it is concluded that the use of machine learning tools for this type of task provides comparable and accurate results, but uses a disproportionate amount of data as a research base, what is important when forming complex regressions and forecasts.
Keywords: Mathematical methods, forecasting, machine learning, transfer of innovative developments, conversion rate of transfer funnel.
|
Olga Pyataeva 
PDF 