Fig 1.
Histograms of the inputs in the current database, (a) cement; (b) blast furnace slag; (c) fly ash; (d) water; (e) superplasticizer; (f) coarse aggregate; (g) fine aggregate; (h) age; (i) early compressive strength.
Fig 2.
Summary of the prediction performance over 100 simulations using different train-to-test ratios and different criteria: (a) R; (b) RMSE; and (c) MAE, where µ denotes the average value, σ denotes the standard deviation, and 25%-75% denotes the values in the range of the first and the third quartiles, respectively.
Fig 3.
Regression graphs for the measured and predicted values of early compressive strength of HPC for (a) the training dataset; (b) testing dataset; and (c) all dataset.
Table 1.
Summary of the statistical measures for the training and testing datasets.
Table 2.
Comparison with literature for prediction of compressive strength of HPC.
Fig 4.
Regression graphs for the measured and predicted values of early compressive strength of HPC for all dataset: (a) ANN; and (b) SVM.
Fig 5.
ICE and PDP curves in function of input variables for: (a) cement; (b) blast furnace slag; (c) fly ash; (d) superplasticizer; and (e) age.
Fig 6.
ICE and PDP curves in function of input variables for (a) water; (b) coarse aggregates; and (c) fine aggregates.
Table 3.
PDP investigation of the compressive strength in function of different inputs and the corresponding effects, rank.
Fig 7.
Two-dimensional PDP curves analysis between cement and other input variables for: (a) BFS; (b) FA; (c) water; (d) superplasticizer; (e) coarse aggregates; (f) fine aggregates; and (g) age. The color scale presents a variation of compressive strength in MPa.