Multidimensional analysis of engineering cost database based on descriptive data mining

Abstract

A vital area of study is cost estimation in building engineering. Traditional models' poor capacity to appropriately capture the changing trend of construction engineering costs leads to low precision in cost forecast for construction engineering. The result is the recommendation of a descriptive data mining-based cost prediction model for construction engineering. The engineering cost database is initially constructed using a number of distinctive indices, and the aberrant data are discovered and filtered using the K-means clustering method. Second, the mathematical model is created, and the LSSVM is used to solve it. The model is then optimized using the PSO method. According to the experimental findings, this model's prediction accuracy is good and its average root-mean-square error (RMSE) for 100 samples is 2.27-e. The estimated value of individual cost in this model is the most accurate when compared to other models, and the predicted difference is just 30 Yuan/m2. Because it can estimate engineering costs with greater accuracy and stability, this finding has important implications for engineering investment choices.