One-Hot Encoding

What is one-hot encoding?

One-hot encoding converts a categorical column into multiple binary columns.

Example:

text

- city = {Pune, Delhi}

becomes:

• city_Pune (0/1)
• city_Delhi (0/1)

One-hot encoding with Pandas

get_dummies

import pandas as pd
 
df = pd.DataFrame({"city": ["Pune", "Delhi", "Pune"], "amount": [100, 200, 150]})
 
encoded = pd.get_dummies(df, columns=["city"], drop_first=False)
print(encoded)

Avoid dummy variable trap (optional)

For some linear models, you can drop one category:

drop_first

encoded = pd.get_dummies(df, columns=["city"], drop_first=True)
print(encoded)

One-hot encoding with scikit-learn

OneHotEncoder

import pandas as pd
from sklearn.preprocessing import OneHotEncoder
 
X = pd.DataFrame({"city": ["Pune", "Delhi", "Pune"]})
 
enc = OneHotEncoder(sparse_output=False, handle_unknown="ignore")
 
arr = enc.fit_transform(X[["city"]])
print(arr)
print(enc.get_feature_names_out(["city"]))

Tips

• High-cardinality categories (thousands of unique values) can explode feature count.
• Consider grouping rare categories into “Other”.

Visualize it

diagram One-hot encoding mermaid

  flowchart LR
  A["city column
(Pune, Delhi, Pune)"] --> B["pd.get_dummies(df, columns=['city'])"]
  B --> C["city_Pune (0/1)"]
  B --> D["city_Delhi (0/1)"]

A single categorical column expands into one binary column per distinct category.

sketch One column becomes many binary columns p5.js Pause Restart

Each row keeps a single 1 marking which category it belongs to; every other column for that row is 0.

From the book

McKinney’s pattern for merging dummies back with the rest of a DataFrame: dummies = pd.get_dummies(df["key"], prefix="key") then df[["data1"]].join(dummies). Passing prefix keeps the new binary columns readable once they’re joined back in — useful if more than one column gets one-hot encoded.

🧪 Try It Yourself

Exercise 1 – One-hot encode with pandas

pch.exerciseTag pch.exerciseTitle

Exercise 2 – Drop the first category

pch.exerciseTag pch.exerciseTitle

Exercise 3 – One-hot encode with scikit-learn

pch.exerciseTag pch.exerciseTitle

Next

For categories that do have a natural order, see Label Encoding instead of one-hot encoding.