Predicting Benefits from Event Logs using Generative AI

Abstract

This research presents the development of a Generative Artificial Intelligence for analyzing and predicting the benefits of event logs prior to Process Mining analysis. The system operates in a local environment, which ensures security and prevents the leakage of sensitive data. It utilizes the gpt-oss:latest model via the Ollama platform and is developed in Python to enhance the efficiency of data exploration and interpretation, and to reduce the time required for event log analysis. The system is capable of automated analysis, ranging from generating preliminary analytical overviews and creating diverse analytical perspectives, to producing in-depth reports with actionable recommendations. The case study utilized real-world data from a hospital’s outpatient department. The results demonstrate that the system effectively performs Data Type Classification and rapidly generates predictive analytics that provide deep insights. This study highlights the potential of Local Large Language Models (LLMs) to make advanced process analysis technologies more accessible to non-expert users, reducing the time and complexity involved in the experimental process.

Keywords: Process Mining, Generative AI, Large Language Models, PredictiveAnalysis, Automated Process Discovery, Local LLMs