A machine learning approach to effective evaluation of project planning across construction projects

This thesis addresses the challenges of planning evaluation in the Architecture, Engineering and
Construction (AEC) sector by developing a mechanism for clustering and labelling planning activities.
The research objectives include the development of a method for clustering planning activities using
unsupervised machine learning and implementing a labelling system for new activities based on these
clusters. Real project applications validate the mechanism’s effectiveness, focusing on identifying
bottlenecks and optimizing resource allocation. The central problem tackled in this thesis is the lack of
an automatic and systematic approach to planning evaluation, particularly concerning historical data,
lead times, and the absence of automation and intelligent computing in planning processes. Addressing
this issue, the thesis explores current planning principles, generic category clustering, category
definition, and the requirements for clustering and labelling activities. The thesis is structured with a
systematic and bibliometric literature review, an explanation of the research methodology,
presentation of results, validation of clusters and labels, and a discussion of findings and potential
future research directions. This research contributes to bridging the gap between traditional and
automated planning methods in the industry, taking a significant step towards the automated
evaluation of project planning within the AEC sector.