SMU Data Science Review
Abstract
This paper aims to optimize the investment and position of a building supply company by using a predictive modeling approach that focuses on single-family housing starts in the United States. This approach aims to guide where capital investments such as truss-plants and lumber yards should be constructed before housing demand in a particular Metropolitan Statistical Area (MSA) rises. By accurately predicting these MSAs, a building supply company could save costs and improve service while establishing a competitive advantage in the market. This study leverages Multi-Linear Regressions, XGBoost Regression, and Artificial Neural Network (Multilayer Perceptron - MLP) algorithms to identify the top 5 MSAs measured in total single-family housing starts (SFH). These MSAs include (1) New York-Newark-Jersey City (2) Dallas-Fort Worth-Arlington, TX, (3) Houston-The Woodlands-Sugar Land, TX, (4) Atlanta-Sandy Springs-Alpharetta, GA, and (5) Phoenix-Mesa-Chandler, AZ. Regarding MSAs that are projected to experience growth, however, the following top MSAs were identified: 1) Knoxville, TN, 2) El Paso, TX, 3) Cleveland-Elyria, OH, 4) San Antonio-New Braunfels, TX, and 5) Tucson, AZ. Based on these results, this study suggests investments in infrastructure and capabilities should be considered in these areas primarily due to the projected growth and demand for single-family residential construction.
Recommended Citation
Barker, Jayson; Ramos, Emil; Rodgers, John; and Shahzad, Saqib
(2021)
"Machine Learning Forecasting of Construction Demand in the USA as Indicated by Economic, Geographic and Demographic Cues,"
SMU Data Science Review: Vol. 5:
No.
2, Article 4.
Available at:
https://scholar.smu.edu/datasciencereview/vol5/iss2/4
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License
