Encoding in Python with pandas
You'll learn how to convert categorical variables into numeric dummy variables using pd.get_dummies() for regression modeling.
The Core Idea
Once you understand dummy variable encoding conceptually, you need a practical tool to actually create those 0/1 columns from categorical data. In Python, pandas provides pd.get_dummies() — a function that automatically transforms categorical columns into model-ready dummy variables.
Think of pd.get_dummies() as a translator: it takes text categories like "Red", "Blue", "Green" and converts them into separate binary columns that regression algorithms can understand.
Basic Usage
import pandas as pd
# Sample data with a categorical variable
df = pd.DataFrame({
'color': ['Red', 'Blue', 'Red', 'Green', 'Blue'],
'price': [100, 150, 110, 130, 145]
})
# Create dummy variables
df_encoded = pd.get_dummies(df, columns=['color'])
This creates new columns: color_Red, color_Blue, color_Green with 0s and 1s.
Avoiding the Dummy Variable Trap
By default, pd.get_dummies() creates one column per category. To avoid perfect multicollinearity, drop one reference category:
df_encoded = pd.get_dummies(df, columns=['color'], drop_first=True)
Now you'll have color_Green and color_Red, with color_Blue as the implied reference.
Practical Considerations
- Use
drop_first=Truefor regression models to prevent the dummy variable trap - Keep the original data separate if you need category names later
- For multiple categorical columns, list them all:
columns=['color', 'size'] - Column names automatically include the original variable name as a prefix
Key Takeaway: pd.get_dummies() with drop_first=True efficiently converts categorical variables into regression-ready dummy variables while automatically handling the reference category.