Regression Analysis is a statistical method for modeling relationships between variables. It predicts outcomes, quantifies impact, and supports decision-making. Two of the most widely used methods are Linear Regression and Logistic Regression.
Regression estimates how a dependent variable (outcome) changes when one or more independent variables (predictors) change. It helps identify patterns, trends, and predictive power.
Linear Regression models the relationship between variables using a straight line. It is used for predicting continuous outcomes.
Formula: Y = β₀ + β₁X + ε
Predicting monthly sales (Y) based on advertising spend (X). If β₁ = 2,000, it means each extra $1 spent on ads increases sales by $2,000.
Logistic Regression is used when the dependent variable is binary (yes/no, churn/not churn). Instead of predicting a number, it predicts the probability
Formula: P(Y=1) = 1 / (1 + e^(−(β₀ + β₁X)))
Predicting if a customer will churn (Y=1) based on usage time and complaint history. The model outputs P(churn) = 0.82 → high risk of churn, so retention offers can be applied.
| Aspect | Linear Regression | Logistic Regression |
|---|---|---|
| Dependent Variable | Continuous (e.g., revenue) | Categorical (e.g., churn yes/no) |
| Output | Numeric values | Probabilities between 0 and 1 |
| Use Case | Forecasting sales, costs | Customer churn, fraud detection |
| Model Form | Straight line | S-shaped curve (sigmoid) |