Özet:

Concrete compressive strength prediction is vital to the durability and safety of structures in the construction industry. An accurate prediction ensures reliability in the design of structures and contributes to the successful completion of structural engineering projects. Furthermore, predictions play an important role in making critical decisions such as material selection and structural safety calculations. Therefore, concrete compressive strength prediction is a fundamental step to ensure quality control and safety in the construction industry. Therefore, this study aims to predict concrete compressive strength with high accuracy. For this purpose, prediction processes were performed with Decision Tree Regression, Support Vector Regression, Linear Regression and Neural Networks on the Concrete Compressive Strength data set obtained from UCI. In the study, a new dataset with 50 features was created by using feature engineering technique on the dataset. Concrete compressive strength predictions were made by reapplying the same methods on the new data set. The predictions made without applying feature engineering and the predictions made on the new data set obtained by applying feature engineering are presented in a controversial and comparative manner. The findings show that feature engineering improves the performance of the machine learning methods used in the study, while the most successful method is obtained from the structure established with machine engineering and neural networks.

Anahtar Kelimeler: