Farouq Sammour, Heba Alkailani, Ghaleb J. Sweis, R. Sweis, Wasan Maaitah, Abdulla Alashkar
{"title":"Forecasting demand in the residential construction industry using machine learning algorithms in Jordan","authors":"Farouq Sammour, Heba Alkailani, Ghaleb J. Sweis, R. Sweis, Wasan Maaitah, Abdulla Alashkar","doi":"10.1108/ci-10-2022-0279","DOIUrl":null,"url":null,"abstract":"\nPurpose\nDemand forecasts are a key component of planning efforts and are crucial for managing core operations. This study aims to evaluate the use of several machine learning (ML) algorithms to forecast demand for residential construction in Jordan.\n\n\nDesign/methodology/approach\nThe identification and selection of variables and ML algorithms that are related to the demand for residential construction are indicated using a literature review. Feature selection was done by using a stepwise backward elimination. The developed algorithm’s accuracy has been demonstrated by comparing the ML predictions with real residual values and compared based on the coefficient of determination.\n\n\nFindings\nNine economic indicators were selected to develop the demand models. Elastic-Net showed the highest accuracy of (0.838) versus artificial neural networkwith an accuracy of (0.727), followed by Eureqa with an accuracy of (0.715) and the Extra Trees with an accuracy of (0.703). According to the results of the best-performing model forecast, Jordan’s 2023 first-quarter demand for residential construction is anticipated to rise by 11.5% from the same quarter of the year 2022.\n\n\nOriginality/value\nThe results of this study extend to the existing body of knowledge through the identification of the most influential variables in the Jordanian residential construction industry. In addition, the models developed will enable users in the fields of construction engineering to make reliable demand forecasts while also assisting in effective financial decision-making.\n","PeriodicalId":45580,"journal":{"name":"Construction Innovation-England","volume":" ","pages":""},"PeriodicalIF":3.1000,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Construction Innovation-England","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1108/ci-10-2022-0279","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"CONSTRUCTION & BUILDING TECHNOLOGY","Score":null,"Total":0}
引用次数: 1
Abstract
Purpose
Demand forecasts are a key component of planning efforts and are crucial for managing core operations. This study aims to evaluate the use of several machine learning (ML) algorithms to forecast demand for residential construction in Jordan.
Design/methodology/approach
The identification and selection of variables and ML algorithms that are related to the demand for residential construction are indicated using a literature review. Feature selection was done by using a stepwise backward elimination. The developed algorithm’s accuracy has been demonstrated by comparing the ML predictions with real residual values and compared based on the coefficient of determination.
Findings
Nine economic indicators were selected to develop the demand models. Elastic-Net showed the highest accuracy of (0.838) versus artificial neural networkwith an accuracy of (0.727), followed by Eureqa with an accuracy of (0.715) and the Extra Trees with an accuracy of (0.703). According to the results of the best-performing model forecast, Jordan’s 2023 first-quarter demand for residential construction is anticipated to rise by 11.5% from the same quarter of the year 2022.
Originality/value
The results of this study extend to the existing body of knowledge through the identification of the most influential variables in the Jordanian residential construction industry. In addition, the models developed will enable users in the fields of construction engineering to make reliable demand forecasts while also assisting in effective financial decision-making.