Logistic regression will comes under supervised learning to solve classification problems.some real time use cases like if customer is good or bad,predicting defaulters, Email spam detection, Fraud detection etc..
Great I have a question why can’t we use linear regression to solve classification problem?
As we seen above the usage of logistic regression we need to predict good(1) or bad(0), spam (1) or not spam(0), so the outcome of the classification problem is binary, in linear regression the outcome is infinity, so we can’t use Linear regression for classification problem.
Before we go to logistic regression, we will have a discussion on probability ,odds,logarithm and exponential .
We will discuss probability and odds with a problem:
what is the probability and odd of picking 3 blue balls out of 2 blue and 1 red.
So there is a 50% of probability of picking black color balls.
Click here to check to check how to calculate the logistic regression.
Data and Machine by viswateja 