Fig. 1. (Color online) Sensitivity of the developed PSO-SVR model on the number of generation keeping the number of population at 200.
Fig. 2. Correlation cross-plot between the estimated and experimental band gap.
Fig. 3. (Color online) Effect of cobalt-sulfur co-doping on the band gap of TiO2 using the developed PSO-SVR model.
Fig. 4. (Color online) Effect of nickel-iodine co-doping on the band gap of TiO2 using the developed PSO-SVR model.
Fig. 5. (Color online) Effect of tungsten doping on the band gap of TiO2 using the developed PSO-SVR model.
Fig. 6. (Color online) Effect of indium doping on the band gap of TiO2 using the developed PSO-SVR model.
Parameter | Lattice parameter a (Å)
| Lattice parameter c (Å)
| Band gap (eV) |
---|
Mean (eV) | 3.783 | 9.498 928 | 2.943 27 | Maximum (eV) | 3.807 | 9.667 | 3.458 | Minimum (eV) | 3.763 | 9.363 | 1.75 | Standard deviation | 0.008 | 0.042 441 | 0.356 839 | Correlation coefficient (%) | 1.42 | −42.80 | |
|
Table 1. Statistical analysis of the dataset.
SVR hyper parameter | Optimum value |
---|
Regularization factor (C)
| 1 | Epsilon (
)
| 0.054 | Kernel option (
)
| 0.0203 | PSO optimum parameters | Optimum value | Number of population | 200 | Number of generation | 50 |
|
Table 2. Optimum model parameters.
Performance evaluation parameters | Training dataset | Testing dataset |
---|
RMSE (eV) | 0.253 | 0.165 | MAE (eV) | 0.163 | 0.125 | CC (%) | 72.56 | 84.13 |
|
Table 3. Performance evaluation parameters and their corresponding values.