Table 2 Novel feature selection algorithms (NFSAs)
From: Predictive modelling of thermal conductivity in single-material nanofluids: a novel approach
Feature selection type | Brief description of the feature selection algorithm | Selected features and their importance | Model performance | |
|---|---|---|---|---|
Novel feature selection algorithm is based on similar skewness and data resemblance | This selects variables that have close to or the same statistical characteristics | TC, DP, VF, NPk, NPd, BFkv, BFv | RMSE (Validation) | 1.74 |
MSE (Validation) | 3.01 | |||
RSQUARED (Validation) | 0.94 | |||
MAE (Validation) | 1.14 | |||
MAE (Test) | 1.01 | |||
MSE (Test) | 2.26 | |||
RMSE (Test) | 1.50 | |||
RSQUARED (Test) | 0.95 | |||
Novel feature selection algorithm is based on different skewness and data resemblance. (The best) | This selects variables that have dissimilar. Statistical characteristics, differing values among neighbours with different response values | TC, DP, VF, NPk, NPmp, BFkv, BFv | RMSE (Validation) | 1.83 |
MSE (Validation) | 3.34 | |||
RSQUARED (Validation) | 0.94 | |||
MAE (Validation) | 1.23 | |||
MAE (Test) | 0.99 | |||
MSE (Test) | 2.14 | |||
RMSE (Test) | 1.46 | |||
RSQUARED (Test) | 0.97 | |||