Logistic Regression and Text Classification

4.13 Interpreting models

1. What does it mean for a classifier to be interpretable, and how can logistic regression be interpreted?
   We can know why it reached its decision. With human-designed features, inspect each feature’s weight magnitude or test its significance (likelihood ratio test, Wald test).
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 4, §4.13, p. 121
2. How is logistic regression used as an analytic tool, and what must be controlled for?
   To test whether a feature (e.g. negation words) is associated with an outcome (negative sentiment, hospital readmission, sales) above and beyond other variables. Potential confounds (genre, year, review length, patient age…) must be included as controls.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 4, §4.13, p. 121