Implementing machine learning methods in Stata

Implementing machine learning methods in Stata

1. What are machine learning algorithms (MLA)?

   • Methods to derive a rule from data, or reduce the dimension of available information
     • Also known as data mining, data science, statistical learning, or statistics.
       • Or econometrics, if you are in my tribe.

2. Fundamental distinction

   • Most MLA are designed to reproduce how a human would classify something, with all inherent biases. No pretension to deep structural parameters or causal inference—but this is changing.

3. Unsupervised MLA: no labels (no outcome data)

   • Clustering: cluster kmeans, kmedians
   • Principal component analysis: pca
   • Latent class analysis: gsem in Stata 15

4. Supervised MLA: labels (outcome y)

   • Bagging e.g. RandomForest
   • Boosting e.g. AdaBoost
   • Neural Networks (NN), Convolutional NN
   • Tree builders such as ID3 (Quinlan, 1986), C4.5 (Quinlan, 1993), CART (Breiman et al., 1984)
   • “Structural” Equation Models e.g. sem, gsem, irt, fmm
   • Support Vector Machines or kernel machines
   • Nonparametric regress e.g. lpoly, npregress
   • Generalized additive models (findit gam), wavelets, splines (mkspline)
   • Shrinkage: lasso, ridge regression, findit lassopack
   • Nonlinear discriminants: discrim knn
   • Regression or linear discriminants: regress, discrim lda