Encoding categorical variables in Data Analysis Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of One-Hot Encoding with pandas

What is the output DataFrame after applying one-hot encoding to the 'Color' column using pandas get_dummies?

Data Analysis Python

import pandas as pd

df = pd.DataFrame({'Color': ['Red', 'Blue', 'Green', 'Blue']})
df_encoded = pd.get_dummies(df, columns=['Color'])
print(df_encoded)

A

   Color_Blue  Color_Green  Color_Red
0           0            0          1
1           1            0          0
2           0            1          0
3           1            0          0

B

   Blue  Green  Red
0     0      0    1
1     1      0    0
2     0      1    0
3     1      0    0

C

   Color_Blue  Color_Green  Color_Red
0           1            0          0
1           0            1          0
2           0            0          1
3           0            1          0

D

   Color_Blue  Color_Green  Color_Red
0           0            1          0
1           1            0          0
2           0            0          1
3           1            0          0

💡 Hint

Remember that get_dummies creates a new column for each category with 1 where the row matches that category.

The get_dummies function creates columns named with the original column name plus the category. Each row has 1 in the column matching its category and 0 elsewhere.

❓ data_output

intermediate

Label Encoding Result

What is the array output after label encoding the 'Fruit' list using sklearn's LabelEncoder?

Data Analysis Python

from sklearn.preprocessing import LabelEncoder

fruits = ['apple', 'banana', 'apple', 'orange', 'banana']
encoder = LabelEncoder()
encoded = encoder.fit_transform(fruits)
print(encoded)

A [0 1 0 2 1]

B [1 2 1 3 2]

C [2 1 2 0 1]

D [0 0 1 2 1]

💡 Hint

LabelEncoder assigns integers starting from 0 in alphabetical order of categories.

The categories sorted alphabetically are 'apple', 'banana', 'orange'. They get labels 0, 1, 2 respectively.

🔧 Debug

advanced

Error in One-Hot Encoding with Unknown Categories

What error will this code raise when transforming new data with unseen categories using sklearn's OneHotEncoder?

Data Analysis Python

from sklearn.preprocessing import OneHotEncoder

encoder = OneHotEncoder(handle_unknown='error')
encoder.fit([['cat'], ['dog']])
encoder.transform([['cat'], ['bird']])

A KeyError: 'bird'

B ValueError: Found unknown categories ['bird'] during transform

C TypeError: unhashable type: 'list'

D No error, output is a sparse matrix

💡 Hint

By default, OneHotEncoder raises an error if it sees categories not seen during fit.

The parameter handle_unknown='error' causes a ValueError when transform sees unseen categories like 'bird'.

🚀 Application

advanced

Choosing Encoding for High Cardinality Feature

You have a categorical feature with 10,000 unique values. Which encoding method is best to reduce memory and avoid too many columns?

A Label encoding

B Frequency encoding

C Binary encoding

D One-hot encoding

💡 Hint

One-hot encoding creates one column per category, which is large here.

Binary encoding reduces dimensionality by encoding categories as binary digits, using fewer columns than one-hot.

🧠 Conceptual

expert

Effect of Label Encoding on Tree-Based Models

Why can label encoding categorical variables be problematic for linear models but usually acceptable for tree-based models?

A Linear models handle missing values better than tree models when using label encoding.

B Tree models require numeric labels, linear models do not.

C Label encoding creates dummy variables that confuse linear models but not tree models.

D Linear models assume numeric order in labels, which can mislead them; tree models split on values without assuming order.

💡 Hint

Think about how models interpret numeric values of categories.

Linear models treat encoded labels as ordered numbers, which can mislead them if categories have no order. Tree models split data based on thresholds and do not assume order.