Logistic Regression is used when the output is categorical (Sure/No, 0/1, True/False). As a substitute of predicting a steady quantity, it predicts the chance of an occasion taking place.
P(Y)=1/1+e−(b0+b1X)
The place:
- P(Y) = Chance of an occasion occurring
- e = Exponential operate
An S-shaped (sigmoid) curve that predicts values between 0 and 1
Instance:
Predicting whether or not an individual has diabetes (Sure = 1, No = 0) primarily based on blood sugar ranges.
If an individual has a blood sugar degree of 90, they won’t have diabetes (0), but when their blood sugar degree is 150, they’re extra more likely to have diabetes (1).
- Spam electronic mail detection (Spam = 1, Not Spam = 0)
- Buyer churn prediction (Will depart = 1, Will keep = 0)
- Fraud detection in banking (Fraud = 1, Not Fraud = 0)
