Open harryyikhchan opened 2 years ago
The unsupervised clustering has been provided to the partner for further analysis. Once they provide feedback in terms of validity of risk labels based on the unsupervised clusters, the features in supervised models can be adjusted accordingly.
Since there is no ground truth for API risk assessment, we should use clustering analysis to check for the vanity of risk labels.