APPLYING MACHINE LEARNING FOR THE PREDICTION OF CONSTRUCTION WASTE OUTPUT GENRATED DURING THE CONSTRUCTION OF RESIDENTIAL PROJECTS.

The construction industry generates a significant amount of waste that can have serious environmental
and economic consequences. To address this problem, it is important to understand the factors that
contribute to the generation of construction waste. This thesis aims to develop a machine learning
prediction model that can help improve understanding of construction waste generation by analysing
construction project variables and to propose suitable ML-algorithms for current contracting companies.
To achieve this goal, several existing machine learning algorithms were applied and tested. The models
were trained on a dataset that included information on construction projects, such as project size,
location, type of construction, and materials used. The performance of each model was evaluated using
metrics such as accuracy, precision, and recall.
The results of the study showed that machine learning algorithms can effectively predict construction
waste generation based on project variables. The ridge regression algorithm had the highest
performance, with an r2 of 0,744. In addition to developing a prediction model, the study explored ways
to improve the input data and use the output effectively. One approach involved to develop a tool to
extract more data from the building information modelling (BIM) environment to create a more
comprehensive dataset. Overall, this study contributes to the efforts to use ML-algorithms to predict
construction waste, which is a first step towards identifying the key factors that contribute to waste
generation and exploring ways to improve data collection and analysis.