Primena mašinskog učenja za procenu cena i količina radova pri izgradnji stambenih i stambeno-poslovnih objekata

Abstract

Ovaj rad se bavi problemom procena potrebnih količina radova, kao i koštanja izgradnje stambenih i stambeno-poslovnih objekata korišćenjem algoritama mašinskog učenja. Osnovni cilj je analiza mogućnosti primene mašinskog učenja za razvoj modela koji će za kratko vreme pružiti dovoljno preciznu preliminarnu procenu potrebnih količina i cena glavnih radova na osnovu malog broja poznatih parametara. Istraživanje je sprovedeno na osnovu podataka o realizovanim projektima izgradnje višeporodičnih stambenih i stambeno-poslovnih objekata koji su izgrađeni u periodu od 2012. do 2020. godine na teritoriji Republike Srbije. U radu je predloženo nekoliko modela za procenu količina i cena pojedinih vrsta radova, kao i ukupne cene građevinskih radova. Rezultati analize su pokazali da se veća tačnost može postići predikcijom količina nego cena pojedinih radova. Razvijeni modeli mogu biti korisni u procesu planiranja troškova i količina potrebnog materijala u ranim fazama razvoja projekta.

This paper deals with the problem of estimating the required quantities of works, as well as the cost of construction of residential and residential-commercial buildings using machine learning algorithms.The main goal is to analyze the possibility of applying machine learning for the development of a model that will in a short time provide a sufficiently precise preliminary estimate of the required quantities and cost of major works based on a small number of known parameters.The research was conducted on the basis of data on realized projects of the construction of multi-family residential and residential-commercialbuildings that were constructed in the period from 2012 to 2020 on the territory of the Republic of Serbia.The paper proposes several models for estimating the quantities and cost of individual types of works, as well as the total price of construction works.The results of the analysis showed that greater accuracy can be achieved by predicting quantities than the cost of individual works.The developed models can be useful in the process of planning the cost and quantities of required material in the early stages of project development.