IMPROVING A RISK ESTIMATING METHOD FOR THE “SMART HOUSE” INFORMATION SYSTEM IT PROJECT

Abstract

The object of this study is the IT project risk management process. The study solves the task to improve the accuracy of risk assessment in IT projects and, in particular, IT projects to design the information system (IS) “Smart House”. Research into this area is mainly focused on the application of machine learning (ML) methods to improve the results of conventional evaluation methods. Issues related to quantitative risk assessment of IT projects to design the “Smart House” IS remain practically unexplored. During the study, it was proposed to use ML methods for preprocessing the raw data. For this purpose, a combined risk assessment method was devised. In this technique, methods of Support Vector Machine and Bayesian networks were applied to process the raw data. The results of their application were used as input data for Monte Carlo simulations. During the software implementation of the devised method, its technological stack was determined. Fragments of the program code are given, which describe the implementation of the basic elements of the combined method. The devised method and its software implementation were used to assess the risk of delay in the implementation of the IT project to design the “Smart House” IS. The evaluation results determine the expected duration of this IT project at 234.5 days, with a deviation range of 226–244 days (with a 95 % confidence interval). The results of a comparative analysis of the obtained estimates with estimates of the same risk obtained using the conventional Monte Carlo method show that the devised method provides higher reliability of forecasts. The application of research findings makes it possible to improve the quality of managing IT projects by increasing the accuracy and reliability of their risk assessments