<aside> 💡
Logistic regression is a statistical method used when we want to model the probability of a binary (e.g., yes/no) outcome based on one or more predictor variables.
</aside>
Unlike linear regression (such as Multiple Linear Regression) , which predicts continuous values, logistic regression specifically handles cases where the dependent variable is categorical and binary in nature
flowchart LR
A(Regression)
A --> B(Linear)
A --> C(Logistic)
C --> D(Binary outcomes)
B --> E(Continuous outcomes)
D --> F["Yes/No"]
D --> G["Score/NoScore"]
D --> H["Live/Die"]
style A fill:#FFD700,stroke:#FFA500,stroke-width:2px
style C fill:#D8BFD8,stroke:#9400D3,stroke-width:1px
style B fill:#ADD8E6,stroke:#4682B4,stroke-width:1px
This technique gets its name from the logistic function (also called the sigmoid function) that transforms any real-valued number into a value between 0 and 1, making it perfect for probability prediction.
The logistic model basically estimates the probability that an instance belongs to a particular category, either 0 or 1.
The power of logistic regression lies in its ability to handle multiple predictor variables, both continuous and categorical, while providing easily interpretable results in terms of odds ratios.
Review each of the follow elements (in order):
To help consolidate your learning, review the revisions questions here:
Presentation as PDF