Collaboration Process Pattern Approach to Improving Teamwork Performance: A Data Mining-Based Methodology

It is well documented in management literature that characteristics of collaboration processes strongly influence team performance in a business environment. However, little work has been done on how specific collaboration process patterns affect teamwork performance, leading to an open issue in collaboration management. To address this research gap, we develop a Collaboration Process Pattern (CPP) approach that analyzes teamwork performance by mining collaboration system logs from open source software development. Our research is novel in three ways. First, our research is fact-driven, as the result is based on teamwork tracking logs. Second, we develop a pattern mining approach based on sequence mining and graph mining. Third, using time-dependent Cox regression, our approach derives business insights from real-world collaboration data that are directly applicable to managerial actions. Our empirical study identifies collaboration patterns that can lead to more efficient teamwork. It also shows that the effects of collaboration patterns vary depending on the types of tasks. These findings are of significant business value since they suggest that managers should carefully prioritize their limited attention on certain types of tasks for intervention.