4 ms·
Yeah, it is a good first benchmark. But view interpretability as separate from accuracy. You can explain black box algorithms just fine these days. Logistic re
by polkapolka 8y ago
Yeah, it is a good first benchmark. But view interpretability as separate from accuracy. You can explain black box algorithms just fine these days.
Logistic regression is high bias low variance. If you were talking about fairness bias, then resistance to bias comes from logreg being too dumb to recognize complex non-linear patterns. Not necessarily a pro.
- bunderbunder 8y agoSorry, I misspoke - will edit. Was talking about resistance to overfitting. Which largely comes from logistic regression's assumption of a linear decision boundary. It's true surprisingly often in classification tasks, and, when it's not, you can usually model it just fine with interaction variables. With an ANN, your easiest defense against overfitting is to have great big heaping piles of training data. That's something that's hard to come by in many interesting situations.
- polkapolka 8y agoAgreed. Logistic regression with poly kernel or good engineering interactions can equal or beat more complex models for a fraction of the budget. All the more power to you if a solid simple logreg model (or even no ML at all) is your first deliverable.
- tomrod 8y agoWould you mind talking to how black box interpretability is becoming well known? I've seen Shapley values used for feature interpretation, but not sure what else is being done.
- polkapolka 8y agoFor an accessible recent overview see: https://christophm.github.io/interpretable-ml-book/ https://christophm.github.io/interpretable-ml-book/