Multidimensional Analysis of Engineering Cost Database Based on Descriptive Data Mining

Abstract

Abstract Cost estimation in construction engineering is a crucial topic of study. Low precision in construction engineering cost prediction results from traditional models' inability to adequately capture the evolving trend of construction engineering costs. As a result, a descriptive data mining-based cost prediction model for construction engineering is suggested. The database of engineering costs is first built using several characteristic indices, and the aberrant data are found and filtered using the K-means clustering approach. Second, the mathematical model is established, and its solution is carried out using the LSSVM. Finally, the model is optimized using the PSO technique. The experimental results demonstrate that, in comparison to the conventional engineering cost prediction models, our model's predicted value of the single cost is the closest to the actual cost of one party, and the predicted difference is all within 30 yuan/m2. This finding has significant implications for engineering investment decisions because it can achieve higher accuracy and stability in engineering cost prediction.