Overview

• Despite the name 'Logistic Regression', it is actually a probabilistic classification model.
• Its' output is restricted to values between 0 and 1 (via use of the logistic sigmoid function).
• Three types of logistic regression are:

Mathematical Background

• A logistic regression essentially takes the value we get from a linear regression and turns it into a probability.

Process of Operation

• Logistic regression finds a decision boundary between the two classes and outputs are assigned to the class of whichever side of the threshold they lie on.
• Distinguishing between more than two classes can be done using the 1 vs all method.

Training

• There are two common methods used for learning the best weights to use in a logistic regression:

Citations

(n.d.). Machine Learning - Logistic Regression - Tutorialspoint. Retrieved from https://www.tutorialspoint.com/machine_learning_with_python/machine_learning_with_python_classification_algorithms_logistic_regression.htm

(n.d.). Introduction to Logistic Regression - Towards Data Science. Retrieved from https://towardsdatascience.com/introduction-to-logistic-regression-66248243c148

(n.d.). Gradient Descent Training With Logistic Regression - Towards Data Science. Retrieved from https://towardsdatascience.com/gradient-descent-training-with-logistic-regression-c5516f5344f7

(n.d.). machine learning - Gradient descent convergence How to decide convergence? - Stack Overflow. Retrieved from https://stackoverflow.com/questions/17289082/gradient-descent-convergence-how-to-decide-convergence