Skip to main content
. 2021 Sep 1;21(17):5888. doi: 10.3390/s21175888

Table 10.

Time duration for feature importance values generation (CWR case). For the embedded techniques (FS layer, AFS and AFS 2), this value is calculated as the difference between the total training time and the training time for a vanilla model without any embedded techniques. For the random forest technique, this value corresponds to the total training time, as there is no way to identify how much time it takes to obtain the desired values.

Approach FS Layer Mutual Info. ReliefF Random Forest AFS AFS 2
Time [s] 89.3 5.2 595.3 10.2 2377.0 3026.5