Mining Daily Work Report Data for Detecting Patterns of Construction Sequences

Abstract

Sequencing construction activities in highway projects is a complex planning process which requires not only considerable knowledge and practical experience of the planner/scheduler about various relevant aspects, such as the activities themselves, construction and procurement processes, and construction methods, but also input from other key members of the project regarding specific constraints and requirements. Moreover, sequencing is an iterative process; the sequence developed in the planning phase is likely to change in the construction phase. Therefore, learning from as-built schedules of past completed projects is needed to improve the planning and scheduling processes for future projects. In current practices, most state Departments of Transportation (DOTs) still mainly rely on schedulers’ experience for schedule development. A data-driven systematic approach is still lacking, although the highway agencies have been spending a significant amount of money, time, and effort to collect various digital data during the construction process. This study aims to leverage historical digital daily work report data available in the DOTs’ database to detect patterns of construction sequences in highway projects. Daily work report data collected from a state DOT were used to conduct a case study that developed a Sequential Pattern Mining algorithm to extract frequent sequential relationships among the activities for one major type of highway projects. Peer-review under responsibility of the scientific committee of the Creative Construction Conference 2019.