I would like to make a flow chart for an ML classifier and make sure that my thinking is correct.
Here is a little about my sample: I have 3 classes and about 160 features. I suspect that some of those features are redundant. My classes are imbalanced. Please let me know if I should provide more info about the sample.
This is my first try on tackling a machine learning problem. Based on checking the functionality of each of the options I use below, I chose it in that specific order, but i might be wrong.
Could someone just let me know if this flow chart is logical, and if there was a problem in any of the step, could you possible tell me why it is not supposed to be there?
- Get the data
- Make it in a readable format and import it
- Split the data into train/test
- Standardize with StandardScalar
- Perform PCA on train data
- Perform K-fold cross-validation on train data
- Apply SMOTE
- Predict (Using SVC)
- Make a confusion matrix
- Perform GridSearchCV
- Perform PermutationImportance