Machine learning for bottleneck analysis and increased flow efficiency : Optimized process flow and performance at GDM

Sammanfattning

The objective of this study was to investigate how AI-driven analytical techniques can be utilized to identify bottlenecks and improve workflow efficiency within data-intensive business environments. The research adopted a combination of traditional process analysis and advanced machine learning methods, including process mining and Graph Neural Networks for anomaly detection. The project followed an iterative methodology: beginning with requirement analysis and theoretical research, advancing through data cleaning and feature engineering. Results revealed that while traditional methods could highlight clear workflow delays and overloaded high performance individuals, machine learning approaches provided deeper insights into complex process interactions with unexplained bottlenecks. The findings demonstrated that AI-driven analysis can provide suggestions of possible bottlenecks without clear explanation to why those attributes are considered as bottlenecks, but the findings still require a human centered approach to validate what actual bottlenecks are.