USE TRAINING NEURAL NETWORKS FOR PREDICTING PRODUCT DEVELOPMENT OF IT PROJECT

Abstract

The state of development of innovations in Ukraine is characterized by an increase in development on the basis of start-up projects with the use as a project product of information systems of varying complexity. The article analyzes the weak survivability of the results of start-up projects. The conclusion on the need to predict the stages of develop ment of IT project products based on the analysis of the processes of interaction of users (customers) with the infor mation system (product). In this article, components of the model of forecasting of IT products development of innova tive start-up projects are considered based on the analysis of formed datasets of the interactions of prospective clients. We offered the algorithm of formation of initial datasets based on Customer Journey Map (CJM), which are the tool of fixing of events of the interaction of clients with the system. Examples of models of analogues of clients' travel maps are given, which are the basis for recording and analyzing interactions. This fact is the basis for the formation of appropri ate data sets of large dimension. As a mechanism for processing big data sets and building strategies for IT products development, it is proposed to use a learning neural network. Mathematical models for further modeling and analysis of the obtained results are built. We used a simple linear regression analysis to model the relationship between a single explanatory variable and a continuous response variable (dependent variable). An exploratory data analysis method was applied to the available data to find repetitive patterns and anomalies. In the course of the research, we construct ed a model of linear regression implementation using the gradient optimisation approach. The linear models of the scikit-learn library for the regression task were also applied, and the stabilisation regression method was implemented. Modelling and analysis of the obtained results were carried out, which showed greater efficiency over the extended life cycle of IT project products.