Selection and comparison of machine learning models

Decision Tree: Credit Approval Model

Reason for selection:

1. High accuracy

2. Easy to understand

3. Government agencies monitor the loan business, because the information needs to be transparent to a certain extent (the decision tree can clearly tell why an applicant's loan was rejected or approved)

 

Support Vector Machines:

advantage:

Disadvantage: representation of black box model

Case: Optical Character Recognition (Image Data Processing). Because SVMs can learn complex patterns without being overly sensitive to noise, optical patterns can be identified with high accuracy. And the only disadvantage of support vector machine: black box, not so important for image processing. If a SVM can distinguish a cat from a dog, it doesn't really matter how it does it.