0

I found the following statement confusing. This statement upends my basic understanding of predictive machine learning.

"By not thinking probabilistically, machine learning advocates frequently utilize classifiers instead of using risk prediction models."

While currently learning about cross-validation, I'm learning to create predictive functions that utilize predictive features. Cross validation then predicts how well the predictive function will work on the testing dataset.

How are "classifiers" and "predictors" not the same things?

desertnaut
  • 57,590
  • 26
  • 140
  • 166
Gray
  • 1,164
  • 1
  • 9
  • 23
  • @prosoitos we have at least three (!) interrelated SE forums on ML/stats/AI, namely [Data Science SE](https://datascience.stackexchange.com/help/on-topic), [Cross Validated](https://stats.stackexchange.com/help/on-topic), and [Artificial Intelligence SE](https://ai.stackexchange.com/help/on-topic) - certainly no need to point such questions to Computer Science SE... – desertnaut Nov 02 '19 at 09:57
  • Apart from not being about *programming* (or `r`), your question is quite unclear (the title doesn't even agree with your text). Please consider posting to Data Science SE **or** Cross Validated (cross-posting is not allowed), after you have clarified the question. – desertnaut Nov 02 '19 at 10:11
  • Very briefly, a classifier is a model (neural network, SVM, decision tree etc), while a predictor is a variable (feature); a classifier produces classifications, but a predictor does *not* produce predictions. The term "prediction" is a generic term, which in classification settings means, well, a classification, while in regression settings means a numeric prediction; to add somewhat to the confusion, a *regressor* is not a regression model, but a predictor in a regression model... – desertnaut Nov 02 '19 at 10:17
  • The quoted excerpt is again unclear out of context, but you may find the discussion in [Predict classes or class probabilities?](https://stackoverflow.com/questions/51367755/predict-classes-or-class-probabilities/51423325#51423325) useful (I can only *guess* that by "classifiers", the author means *hard classifiers*, but again one cannot be sure out of context). – desertnaut Nov 02 '19 at 10:21

0 Answers0