To determine which classifier(s) to use in the final software.
Goals
Answer the following:
What classification approaches perform best in general on the training data?
Are certain classification problems better for some classifiers than others?
What benefits, if any, are there to combining multiple classifiers?
Approach
Analysis should probably be done in a jupyter notebook (or RMarkdown). Some important steps to take:
Research various classifiers previously used for similar problems (Interaction with #3 )
Try different classifiers and evaluate their accuracy on simple (e.g., predicting male vs female) and complex classification problems (e.g. predicting tissue type, cancer type, etc).
Output
Figures showing the results of benchmarking (e.g., ROC-AUC curves) & relevant methods. Selection of model(s) for optimization.
Purpose
To determine which classifier(s) to use in the final software.
Goals
Answer the following:
Approach
Analysis should probably be done in a jupyter notebook (or RMarkdown). Some important steps to take:
Output
Figures showing the results of benchmarking (e.g., ROC-AUC curves) & relevant methods. Selection of model(s) for optimization.
Depends upon
Depends upon #1 #2
Interacts with #3