Logistic Regression
ELI5 - The Vibe Check
Despite the name, logistic regression isn't about predicting a number, it's about predicting yes or no: spam or not spam, will churn or won't. It takes the same idea as linear regression but squashes the output through a curve so it always comes out as a probability between 0 and 1, like a 73% chance this email is spam.
Real Talk
Logistic regression is a classification algorithm that models the probability of a binary, or multiclass, outcome by applying a sigmoid function to a linear combination of input features, producing outputs bounded between 0 and 1. Despite predating deep learning by decades, it remains a fast, interpretable, and hard-to-beat baseline for many classification problems.
When You'll Hear This
"Before we build a neural net, let's see how far logistic regression gets us." / "Logistic regression is still winning on this dataset, the extra complexity of the ensemble isn't paying off."
Related Terms
Classification
Classification is teaching an AI to sort things into categories. Is this email spam or not? Is this image a cat, dog, or bird?
Confusion Matrix
A confusion matrix is a little table that stops your model from lying to you with a single accuracy number.
Feature Engineering
Feature engineering is the unglamorous, hugely important work of turning raw, messy real-world data into the specific columns a model can actually learn fr...
Linear Regression
Linear regression is drawing the single straightest line through a scatter of dots that comes closest to all of them at once, then using that line to guess...