Let's assume we have a dataset containing information about whether individuals purchased a product (Yes or No) based on their age and salary. The goal is to predict whether a person will purchase the product based on these features.
| Age | Salary | Purchased |
|---|---|---|
| 22 | 50000 | No |
| 25 | 60000 | No |
| 47 | 150000 | Yes |
| 52 | 200000 | Yes |
| 46 | 90000 | Yes |
| 56 | 160000 | Yes |
| 26 | 80000 | No |
| 27 | 58000 | No |
| 48 | 140000 | Yes |
| 50 | 135000 | Yes |
In logistic regression, the probability p of the target variable (Purchased) being 1 (Yes) is modeled as:
$p = \frac{e^{(\beta_0 + \beta_1 \text{Age} + \beta_2 \text{Salary})}}{1 + e^{(\beta_0 + \beta_1 \text{Age} + \beta_2 \text{Salary})}}$
This is called a sigmoid function (not softmax) and you can read more about it here