Machine Learning for Weekly Work Plan Generation and Enhanced Constraint Predication in Last Planner System
Last planner system (LPS) is a popular lean tool being used by the construction industry across the world. The LPS consists of systematic approach involving: 1) milestone schedule, 2) phase schedule, 3) look ahead schedule and 4) weekly work plan. The construction professionals take a lead in facilitation of LPS in a construction organization / project. With the advent of machine learning, the construction management community and more specifically lean community is exploring ways to harness power of machine learning for facilitating last planner system. The efficiency in the process of constraint identification and preparation of weekly work plan is currently entirely dependent on the proficiency of planner and lean champion involved in a construction project. Machine learning can aid this process by predicting the constraints and suggesting a look ahead plan. The existing literature studies have highlighted paucity of structured data from construction projects and contracting organizations, which can 2 be used for training and testing of machine learning algorithms. Hence, there is a need for synthetic data for improving the machine learning models in the context of last planner system.. Further to this, the nature of constraints and activities involved in the last planner system is influenced by the sector of construction project such as real estate, transport and so on. Therefore, the sectoral focus is necessary. In this context, this DRP focuses on use of synthetic data for enhanced constraint prediction in last planner system and also explore opportunity for generation of weekly work plan using machine learning.
