Mining big data to create a tool for empirical observation of continuous safety improvement in a construction company - A progressive case study in the lean environment
In any iterative process, without a system of measurement, controlled improvement cannot be recorded. This is especially true in the construction industry, where error occurs, often with fatal repercussions. As part of a process to facilitate the establishment of this metric, an entirely new application was created. The goal of this application is to measure the causal factors that lead to incidents, which will allow the user and administration to track the circumstances and types of incidents. This enables the company to focus on these problem areas and improve through training. By analyzing these incident trends over time, the company can conclude the following: if training reduces the total number of incidents for a given category (identified through these trends), then the corrective action is working. If not, the team must then redefine the problem, which is part of the aforementioned iterative process. The purpose of this study was to identify a viable metric that captures safety practice improvement over time, and verify that the company’s records indicate a correlation between quantity of incidents recorded and man-hours of exposure decreasing over time. Live server data was provided and a series of queries were performed on relevant tables. These result sets were then placed into a database created by the researcher and manipulated to display trend lines representing incident rates over time, as well as specifically identifying a metric of incident count per month/man-hours per month (companywide). Descriptive statistics were performed, with results indicating that although the reporting process itself was becoming standardized and the latter half of the trend chart showed comparable numbers, there was simply not enough reported data as of yet to provide conclusive evidence on the impact of lean practices as it relates to incidence quantity. It is the researcher’s belief, however, that the data suggests an inverse relationship, as the quantity of human reported incidents had increased as a result of standardized practices and effectively captured more instances that may have likely previously gone unrecorded.
Jenkins, Purdue University.
Information Technology|Civil engineering|Vocational education
Off-Campus Purdue Users:
To access this dissertation, please log in to our