Linguistic model of heuristic planning in IT projects
Abstract
Linguistic model of heuristic planning in IT projects
Incoming article date: 20.03.2018One of the key points of planning individual iterations in IT projects is the forecasting of the so-called focus factor. More specifically, it is the concentration of the workforce and the team's efforts to perform tasks. The purpose of this work was to clarify the previously obtained results on the choice of optimal methods of such prediction. This used another criterion for the quality of the forecast, not a mistake in planning the performance of the iteration, as before, and the inaccuracy of the prediction of the focus factor. The analysis carried out according to the new criterion fully confirms all the results obtained earlier. Among the considered mathematical methods of forecasting for planning in iterations of IT-projects the most suitable is extrapolation by moving average (both arithmetic and geometric) for the previous 1-4 periods. The error of the focus factor forecast for them is about 4.7%. Another aim of this work was the formalization of the intuitive methods of planning undertaken by specific teams of developers of projects. On this basis, within the framework of the theory of fuzzy sets with the use of linguistic variables, a heuristic forecasting model is developed. Numerical experiments on the basis of specific production data have shown for it a higher accuracy of the forecast than for extrapolation. The error was about 3% against 4.7%. The paper also presents a combined model using elements of both the heuristic approach and the above-mentioned extrapolation. It showed even higher accuracy of the forecast, with a mean square error of about 1.5%. The results obtained in this work, in our opinion, completely solve the problem of the optimal method of planning individual iterations in the implementation of IT projects.
Keywords: management, IT-project, iterative nature, iteration planning, forecast, focus factor, moving average, standard error, linguistic variables, fuzzy analysis, heuristic model