The Process Discovery Approaches for Decision Making in Sales Activities

Abstract

In the decision-making of sales activities, the means for greatly increasing the efficiency of sales activities are required by eliminating individuals' factors such as experience and intuition. For this purpose, we are developing a business decision support system using a machine learning model. To establish a process discovery method that extracts regularity from the decision-making, it is necessary to learn the process of sales activities with high order acceptance probability. In this paper, we introduce novel two preprocessing techniques of process mining: an activity estimation based on unstructured data such as daily business reports and a process estimation for stochastically expressing the regularity in an atypical process, and finally find out effective learning algorithms. Especially in process estimation, comparison of plural algorithms reveals the superiority of Hidden Markov Model. Though Hidden Markov Model has an issue to need a large amount of daily business report data including time series, by automatically generating activity data between salespersons and customers with a simulator, it is realized to increase the number of training data. As a result, it is confirmed that the new process discovery method presented in this paper is effective for discovering atypical decision-making process.