Back to search

Clemson University

Assessing the Efficacy: Statistical Models vs. Machine Learning (ML) Approaches for Prediction Modeling In the Construction Industry

Abstract

dc:description.abstract

The construction industry has witnessed a significant surge in the daily volume of objective data (i.e., precise data sources representing actual project progress) accumulated across a project's lifecycle. This abundance of data presents an opportunity to extract valuable organizational insights and potential remedies for project management issues. The construction technology landscape is gradually evolving towards integrated software platforms to meet customer needs more effectively. However, the construction sector lacks comprehensive predictive analytics solutions for projects or industry-wide applications - a significant portion of descriptive analytics tools rely on trade association surveys or dashboards constructed from collected company data, suffering from infrequent updates or limited detail. Machine learning (ML) has been experiencing increased adoption within the construction sector. This technology is bringing transformations across various aspects of construction project management, such as risk assessment and mitigation, safety management on construction sites, cost estimation and forecasting, schedule management, and the prediction of building energy demand. The study's research objectives are to assess the efficacy of statistical models vis-à-vis ML-based approaches for prediction modeling in the construction industry by the following: (1) MEASURE the outcomes of the “customer satisfaction” (for the construction coatings sector) prediction model for both ST and ML approaches. (2) Compare ST vis-à-vis ML-based models predicting customer satisfaction in the construction coatings sector for a non-parametric dataset with limited dimensions. (3) DEVELOP a norm for handling non-parametric data with limited dimensions for the construction coatings sector. The norms can assist in decision-making for selecting a method for prediction – statistical or machine learning based on the nature of the dataset, nature of independent variables, nature of factors contributing to the outcome, goal of analysis, and prediction, specifically non-parametric dataset with limited dimensions. These lessons learned can yield substantial advantages for businesses by improving the performance measures of construction projects—a critical measure of project success. This shall benefit industry data analysts conducting project feasibility studies, giving them insights for improved budget allocation and portfolio management plans.

Degree

thesis:*
Name thesis:degree_name
Master of Construction Science and Management (MCSM)
Level thesis:degree_level
Thesis
Discipline thesis:degree_discipline
Construction Science and Management

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Garg, Stuti
Contributors dc:contributor
  • Dr. Vivek Sharma
  • Dr. Dhaval Gajjar
  • Dr. N. Mike Jackson
  • Dr. Jason D. Lucas
  • Dr. Jong Han Yoon

Subjects

dc:subject × 6

Identifiers

dc:identifier.*
Repository record dc:identifier
https://open.clemson.edu/all_theses/4362

Chain of custody

source
Harvested from
Clemson University
Base URL
open.clemson.edu/do/oai/
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Garg, Stuti. Assessing the Efficacy: Statistical Models vs. Machine Learning (ML) Approaches for Prediction Modeling In the Construction Industry. Thesis thesis, https://open.clemson.edu/all_theses/4362