Разработка алгоритма адаптивной оценки продолжительности выполнения проектных задач

The article deals with the problem of underestimation or overestimation of time required for any particular task in a company to be completed. Tasks were exported from the help desk software. The main reasons of incorrect time evaluation are determined: tasks complete time is evaluated only by the head of the company, and tasks project and employee who perform the task are not considered. Objective of the study. Increasing the accuracy of estimating task duration with considering of selected employee and task project. Materials and methods of research. A model that describes the process of setting tasks and working on tasks is developed for increasing task duration estimation. Controllable, uncontrollable external factors and internal parameters of the system are defined for the right problem definition and criteria for estimating the duration of the task is drafted. The analysis of the initial situation is carried out, the computed estimation accuracy is 1.26. 3 categories and 33 types of tasks are identified during the analysis. The experts are grouped to estimate the execution time of tasks of different types. The coefficients of each expert’s competences are calculated. An estimate of the execution time for each type of task is calculated, with considering the weight of the relevance of each expert. The conclusion that expert assessment do not fully solve the problem is made. Adaptive algorithm of estimating project task duration is proposed, which is based on the time calculated by experts, with considering the personal coefficients of employee who performs the task. A three-dimensional weighting matrix, which consider the project, the type of task, and the employee who will perform the task, is created. The corresponding matrix coefficients are regularly updated after the task is completed, which makes it possible to take into account the individual characteristics of each employee and track the time progress of performing tasks of the same type. Due to the specifics of work in the company, when calculating the coefficients, the “forgetting” parameter is introduced, which allows to increase the planned time for completing the task due to last time when employee completed task of this project. Results. The algorithm is implemented in the company help desk software. The system with the algorithm is trying to adapt to the personal characteristics of each employee and to estimate more accurately personally his time to complete any task. The suggested adaptive algorithm worked for 3 months, during which more than 2000 tasks were completed. The calculated estimation accuracy is 0.74. The improvement in accuracy is 41%. Conclusion. The algorithm and developed module for company help desk software made it possible to significantly increase the accuracy of determining the time to complete each task, the time for task evaluation is reduced to several seconds due to the possibility of evaluation by any of the employees, and the task decomposition process is improved. In the future, it is planned to add a text recognition system to automatically determine the type of task.