Compressive Strength Prediction of Nano-Modified Concrete.
Description
Original database presents the datasets compiled from multiple literature sources, comprising a total of 94 data points.Figure 5 presents the correlation coefficients among the nine variables, clearly indicating the specific instances of positive or negative correlation between each pair of variables as well as the varying intensities of their relationships. Figure 6 displays the model performance evaluation metrics for each fold in the 10-fold cross-validation of the four models. Figure 7 provides numerical values for five evaluation indicators based on the prediction results of the four machine learning models on the test set. Figure 8 demonstrates the determination coefficients of each model by comparing the outcomes of the respective algorithms' computer code runs with the actual values. Figure 9 presents the specific predicted values, allowing for an assessment of whether the models tend to underestimate or overestimate the actual values through a comparison of the differences between them.
Files
Steps to reproduce
The process involves collecting and integrating experimental data from published literature to obtain eight primary materials that constitute nano-modified concrete materials, as well as the concrete strength measured through experimental tests under different material ratios. By writing and running various machine learning algorithm codes in a Python environment, specific data on the prediction of concrete strength by different algorithms can be obtained, along with performance evaluation metrics for each model.