Business Analytics > Predictive Modelling > Logistic Regression

Logistic Regression

Linear regression or multiple linear regression we predict the value of dependent variable which is continuous and with the help of independent variables we receive a continuous prediction. Instead of predicting a continuous value logistic regression predicts a probability between 0 and 1. Then this probability is used to classify dependent variable as :

• If probability ≥ 0.5 → Class = 1 (Yes)
• If probability < 0.5 → Class = 0 (No)

Logistic regression is a statistical method used to predict a categorical outcome, usually yes/no (0 or 1), based on one or more independent variables. It works well for binary classification problems and widely used in business, medicine, and social science

 

Example: Let, we have to predict whether a customer will buy a product or not based on Income, Age, Advertising exposure. The Output are like

• 1 → Will buy
• 0 → Will not buy

This approach uses uses a logistic (sigmoid) function to convert any value into a probability:

Picture shows that output is always between 0 and 1 and Produces an S-shaped curve. It also indicates that the

• Higher input values → Higher probability of class = 1
• Lower input values → Lower probability e. class=2

 

Prev

Regression assumptions that have to Test & keep in mind

Next

Decision trees and classification models

Feedback

Submit