A Variable-Clustering-Based Feature Selection to Improve Positive and Negative Discrimination of P53 Protein in Colorectal Cancer Patients

Algorithm 1

Pseudocode of IV_Cluster Methodology.

Input: Datasets D, Feature sets F

Output: K features

1: Use the optimal decision tree algorithm to do variable binning on the feature set.

2: Do WOE coding for variable binning results.

3: Map the dataset D to a new dataset D1 through the WOE encoding of the variable.

4: Use the formula (see 2) to calculate the variable IV value.

5: According to the required K feature sets, use the clustering algorithm to do K-cluster variable clustering on D1.

6: The variable with the largest IV value is selected from the K clusters to form K feature subsets.