A mathematical model for project cost prediction combining multiple algorithms

As a key link in engineering construction, reasonable evaluation of engineering costs can effectively control the budget and save costs. Therefore, the reliability of the engineering cost estimation will directly affect the economic status of the whole project. However, traditional prediction models are based on a single machine learning method, which is not generalised enough and has a low accuracy. In view of this, a mathematical model for engineering cost prediction is constructed by combining a random forest algorithm, the ridge regression algorithm and the extreme gradient boosting algorithm to obtain a prediction model with higher generalisation and accuracy and to evaluate the cost of engineering projects reasonably and scientifically. The average relative error between predicted and actual values is only 0.872%. The root mean square error and average percentage error of the fusion model are relatively small. The superiority of the proposed mathematical model of prediction costs is verified, and the model possesses a certain application value in construction engineering, providing practical reference and guidance for engineering cost prediction.