A Tool to Support Propensity Score Weighting for Enhanced Causal Inference in Business Processes

Effectively evaluating the impact of process interventions on business outcomes is crucial for assessing the effectiveness and return on investment of process improvement initiatives. However, this task is challenging due to the complex interplay of factors influencing process execution and performance. This paper presents a comprehensive and versatile tool that combines propensity score weighting and event logs to enhance causal inference in business processes. Propensity score weighting balances the treatment and control groups based on their observed characteristics, mitigating bias and improving the precision of causal estimates. Event logs are the input source of process mining methods, which enable the analysis and understanding of how a process works. Our tool assists practitioners in selecting the most suitable weighting method, assessing treatment-control group balance, and evaluating covariate balance before and after adjustments. We apply the approach and tool to a synthetic dataset, demonstrating their effectiveness and illustrating key insights gleaned from the analysis. We discuss the implications and benefits of this approach for advancing causal inference in business processes, alongside limitations and potential future developments for the tool.