PSI - Issue 70

Ch. Vihas et al. / Procedia Structural Integrity 70 (2025) 461–468

463

2.1 Data Collection & Preprocessing

The process of gathering data from available sources is called data collection. For this project, the dataset includes parameters such as cement, water, and aggregate quantities, along with material properties gathered from various available literature sources. A total of 560 values, including 13 features, were collected from the literature. The 12 input features are Cement, specific gravity (SG) of cement, Water (WA), GGBS, specific gravity (SG) of GGBS, Coarse Aggregate (CA), specific gravity (SG) of CA, Fine Aggregate (FA), specific gravity (SG) of FA, Zone of FA, Super Plasticizer (SP) dosage, and age to predict the output feature Compressive strength (MPa). After data collection, data preprocessing takes place. It involves cleaning and transforming the data in a format suitable for training and analysis. The collected data is arranged in the spreadsheet before the data division process. In the preliminary stage the collected data set is classified into three categories as 70:15:15, 80:10:10, 90:5:5, for training, testing and validation. Based on the performance metrics related to training it is decided to use first 80% of the data for training the model, the second 10% for the validation of data, and the rest 10% for the testing the model on unseen data for further studies. After the data preparation, the correlation heatmap was developed. The Correlation heatmap is a two dimensional representation of the relationship among the dimensions. The features correlation was calculated using Karl Pearson's coefficient. The Pearson coefficient ranges from +1 to -1. Where -1 indicates perfect negative correlation and +1 is perfect positive correlation, 0 represents no correlation among the features. From Figure 1, we can conclude that the compressive strength of the concrete is positively correlated to the number of days of curing and the amount of GGBS. The higher age of concrete results in more hydration and improved strength. Similarly, the inclusion of GGBS results in a denser CSH structure. The compressive strength is negatively correlated to the quantity of coarse aggregate, as coarse aggregate increases porosity in the concrete. 2.1.1 Correlation Heatmap

Fig. 2. Heatmap representing the Correlation among input features