Please use this identifier to cite or link to this item: https://scholarbank.nus.edu.sg/handle/10635/222755
Title: TEMPORAL ANALYSIS OF CONSTRUCTION DATA TO IMPROVE SAFETY
Authors: CAO HOUCHEN
Keywords: Building
PFM
Project and Facilities Management
Goh Yang Miang
2017/2018 PFM
Time Series
Temporal
Data Mining
Construction Safety
Leading Indicators
Accidents
Forecasting
Issue Date: 18-Jun-2018
Citation: CAO HOUCHEN (2018-06-18). TEMPORAL ANALYSIS OF CONSTRUCTION DATA TO IMPROVE SAFETY. ScholarBank@NUS Repository.
Abstract: The construction industry is a sector which churns out huge amounts of data on a daily basis. However, many studies have not fully explored the usefulness of such large amounts of information to improve the safety aspects of construction. Till date, the construction industry continues to be a prominent contributor to workplace injury rates and fatalities. Thus, it is of paramount importance to grasp the information imbedded in these construction data and conduct data analysis to resolve related safety issues. This study deviates from past literature by applying temporal techniques onto construction data to improve its safety. Through time series analysis, this study aims to identify the leading indicators of construction accidents from a given project’s dataset. Furthermore, it also attempts to conduct forecasts on future occurrences of accidents using various time series models. The aforementioned dataset is obtained from a Singapore BCA grade A1 construction company, containing past information regarding safety inspection score, accident cases and project related data collected from 2008 to 2015. From the entire dataset obtained, five projects with complete and sufficient data were selected. Each of these project’s data contains twenty-three potential leading indicators of future accidents. Through the proposed systematic data analysis process, final leading indicators were identified for each of the five projects. Subsequently, these finalised leading indicators were used for forecasting through three different time series models. The Vector Error Correction (VEC) model has shown to be the best model, displaying the lowest root mean squared error (RMSE) value for three out of the total five projects that underwent analysis. The accuracy of VEC model and the entire leading indicator identification process are also in conjunction with the results of similar works found in the current literature. It is recommended that construction companies can utilise the proposed data analysis process to improve on its existing safety issues. Aside from converging solely onto individual project’s leading indicators, this process is also able to identify underlying company-wide safety issues when applied onto a range of datasets with different project natures. Lastly, this research can possibly serve as a preliminary basis that will guide future time series applications in construction safety.
URI: https://scholarbank.nus.edu.sg/handle/10635/222755
Appears in Collections:Bachelor's Theses

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
Cao Houchen 2017-2018.pdf2.61 MBAdobe PDF

RESTRICTED

NoneLog In

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.