Predicting construction projects’ costs and durations with machine learning – A case study from Finland

No Thumbnail Available

URL

Journal Title

Journal ISSN

Volume Title

School of Business | Master's thesis

Date

2024

Major/Subject

Mcode

Degree programme

Information and Service Management (ISM)

Language

en

Pages

57

Series

Abstract

This thesis studies predicting construction projects’ costs and durations with machine learning. Due to the complexity of construction projects, it is a common phenomenon that the projects suffer from cost and schedule overruns. The fourth industrial revolution and the development of computational power have made applying artificial intelligence to the issues of the construction industry more accessible and desirable than ever. Multiple machine learning models were applied on 112 residential apartment buildings’ data consisting of the buildings’ key attributes. Data on 54 attributes was collected from the case company’s database and the apartment buildings’ marketing materials. For the prediction of the actual cost of construction, support vector regression and linear regression had the highest accuracy with a correlation coefficient of 0.99. Also, for predicting the actual duration of construction, support vector regression and linear regression had again the highest accuracy, yet with slightly lower correlation coefficients of 0.94 and 0.93. Moreover, the best prediction power on both costs and duration was on the datasets with attributes selected by correlation-based feature selection. The attributes used for prediction were able to explain most of the variance in the model. Yet, an average error of 1,06M € on costs and 36 days on durations were received with the best prediction models

Description

Thesis advisor

Liu, Yong

Keywords

machine learning, construction project, cost estimation, duration estimation, prediction

Other note

Citation