Skip to main content
. 2023 Jun 30;3(5):308–318. doi: 10.1021/acsenvironau.2c00072

Figure 2.

Figure 2

Soft sensor development framework was comprised by the following five steps: (i) statistical analysis and dataset preprocessing; (ii) model input selection using the Pearson’s correlation coefficient and Recursive Feature Elimination method (RFE); (iii) model development, including cross-validation for metrics tuning and model calibration; (iv) model testing for ML algorithms to compare the prediction accuracy; and (v) final soft sensor models for COD, TSS, and E. coli prediction.