Fault Identification of UHVDC Transmission Based on DF-AD and Ensemble Learning

Table 3

Pseudo-code of RF-RFECV.

---

Algorithm 1: RF-RFECV

Input: data set samples , and classes labels

Output: optimal feature subset

(1): for in 1 : 5 do

(2):  split the dataset samples into 5 groups, select the as test data, the RF model was trained by the other four groups

(3):  train the random forest model

(4):  calculate the accuracy of RF

(5):  evaluate the MDI and remove the least import feature, update feature subset

(6):  repeat step 4 to step 5 until the quantity of features of dataset is 0

(7): end

(8): determine the optimal feature subset with highest accuracy of CV

---