Questions about ML classifiers

[emilyemchen]

Hi @dllussier!
My team and I have come up with a variety of classifier models from this article and I was wondering if you had some suggestions about which ones would be good ones to start with:

NON-LINEAR

• Nearest Neighbors (K-NN) (Cover and Hart, 1967) with K=1 and Euclidean distance metric
• Gaussian Naïve Bayes (GNB)
• Random Forests Classifier (RF) (Breiman, 2001) @anproulx
• Decision trees

LINEAR (sparse l_1 regularization)

• Support Vector Classification (SVC)
• Logistic Regression (Hastie et al., 2009)

NON-SPARSE LINEAR (l_2 regularization)

• Ridge classification
• SVC
• Logistic regression

We are thinking of starting with supervised learning and then perhaps branching out to unsupervised learning if we have time.

[dllussier]

The model you select will depend on what problem you are trying to solve or the question you are interested in and your data. Scikit learn has a nice cheat sheet that can show you how some of the options might be used: https://scikit-learn.org/stable/tutorial/machine_learning_map/index.html