Generalized linear model-based data analytic approach for construction equipment management

The utilisation of equipment on construction sites can be challenging as it is expensive and bulky with difficulties on flexible dispatching. It is therefore essential that equipment be managed efficiently and effectively. Equipment idleness can reflect the utilization efficiency, but there is limited research on quantitative analysis. The aim of this paper is to determine and examine the influencing factors of equipment idleness and predict the possible idleness considering the construction schedule accordingly. Both data preprocessing and data analysis are considered. Site diary data from a real-life project in Hong Kong was used in this research. Firstly, the semi -structured data is processed into a structured schema for analysis. Secondly, a model is proposed to find po-tential correlations and internal insights from the processed data. The number of idle equipment is used as the response variable to establish a multiple generalized linear model (GLM) with negative binomial regression. Location, project stage, activity, and labour are considered as possible independent variables. Three hypotheses are proposed and the model that considers the interaction between the number of labour and the number of activities has the highest fitting accuracy. Model validation shows that the project manager can make timely scheduling adjustments with the corresponding predictive results.