Approach for individual task completion time prediction in business processes

To manage and schedule human resources effectively in business processes, an approach for individual task completion time prediction was proposed. In this method, the status and capability of resources were modeled and evaluated based on the analysis of historical logs. The algorithm of Support Vector Regression (SVR) was adopted, and its advantages of "soft margin" and various kernel functions were developed to realize the real-time prediction for task completion time through the process of data integration, data standardization, model training, parameter optimization and so on, which had considered the current status and the capacity of specific resources. The problems of dealing with sparse data sets were also taken into consideration. The experiment result revealed that the prediction reached a higher accuracy with sufficient logs according to the approach, which would provide real time decisions for the process of resource management and task assignment, especially in the case of high fluctuation of completion time. © 2019, Editorial Department of CIMS. All right reserved.