Table 5 The four feature combinations
From: Predicting blood pressure from physiological index data using the SVR algorithm
Number | Combination | Reason |
|---|---|---|
1 | PTT, HR, PPG, aVF, aVR, aVL, I, II, III, SpO2 | Use all the features |
2 | PTT, HR, PPG, aVF, SpO2 | Because of a significant correlation between the fourth feature (aVF) and the fifth and sixth features (III, II) of 0.9778 and 0.9808, respectively, the fifth selected feature is SpO2, which is ranked seventh and is not correlated with the first four features. |
3 | PTT, HR, PPG, aVF | Select the 4 features with the highest dependence on the target value. |
4 | PTT, HR, PPG | Select the 3 features with the highest dependence on the target value. |