Is Logistic Regression Regression? (datascienceconfidential.github.io)

In a recent exploration of the terminology surrounding logistic regression, machine learning engineer Richard Vale argues against the notion that it does not belong to the category of regression models. The debate stems from confusion in the data science community, where logistic regression is often classified purely as a classification algorithm due to its categorical output. However, Vale clarifies that logistic regression predicts probabilities (i.e., the likelihood of a data point belonging to class 1), which aligns it more closely with the traditional concept of regression that estimates the expected value of a dependent variable given certain features. This distinction is significant for the AI/ML community as it impacts how algorithms are categorized and understood. By emphasizing that logistic regression functions to predict probabilities, rather than directly classifying data points, Vale highlights a broader conceptual misalignment in predictive modeling terminology. He argues for a redefinition of regression, proposing that any model predicting the expectation of a response variable (like logistic regression) should indeed be labeled as regression, while models producing definite class predictions should be termed classifiers. This nuanced understanding can foster clearer communication and more accurate application of statistical methods in machine learning.