Can be used as a classifier → Training a Binary Classifier
Logistic regression: classify sth. discrete → better name would have been Logistic Classification. A fancy animation can be found here and here.

<aside> 💡 Anmerkung: Polynomial Feautures können aus einem nichtlinearen Problem ein lineares Problem machen. So lässt es sich dann auch mit Logistic Regression lösen. Beispiel hier. Siehe auch Winning with Simple, even Linear, Models

</aside>

Like Linear Regression, Logistic Regression has metrics similar to R^2 and is also calculates p-values.

Main idea of Logistic Regression

Logistic Regression - Code

Winning with Simple, even Linear, Models

Logistic Regression Plots

Logistic Regression vs. Linear Regression

The key difference between linear regression and logistic regression lies in their purpose and output:

Linear regression predicts continuous numerical values. It models the relationship between input variables and a continuous dependent variable by fitting a linear equation to the data. The output is a number that can take any value on a continuous scale (like predicting house prices, temperature, or height).

Logistic regression predicts probability values (between 0 and 1) for binary outcomes. It models the probability that an input belongs to a particular class by using the logistic function to transform its output into a probability value. The output is typically interpreted as classification into discrete categories (like email spam/not spam, customer will buy/not buy, disease present/absent).

Other important differences include:

The mathematical function: Linear regression uses a linear function, while logistic regression uses the sigmoid (logistic) function
The loss function: Linear regression typically uses mean squared error, while logistic regression uses log loss (cross-entropy)