ColumnTransformer for mixed types in ML Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of ColumnTransformer with numeric and categorical features

What is the shape of the transformed data after applying this ColumnTransformer?

ML Python

from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import StandardScaler, OneHotEncoder
import numpy as np

X = np.array([[1, 'red'], [2, 'blue'], [3, 'green']])

ct = ColumnTransformer(
    transformers=[
        ('num', StandardScaler(), [0]),
        ('cat', OneHotEncoder(), [1])
    ]
)

X_transformed = ct.fit_transform(X)
print(X_transformed.shape)

A (3, 4)

B (3, 3)

C (3, 5)

D (3, 2)

💡 Hint

Remember that OneHotEncoder creates one column per unique category.

The numeric column is scaled to 1 feature. The categorical column has 3 unique values, so OneHotEncoder creates 3 columns. Total columns = 1 + 3 = 4. Rows remain 3.

❓ Model Choice

intermediate

Choosing transformers for mixed data types

You have a dataset with numeric, categorical, and text columns. Which set of transformers is best to use in a ColumnTransformer?

A RobustScaler for numeric, OrdinalEncoder for categorical, HashingVectorizer for text

B StandardScaler for numeric, OneHotEncoder for categorical, TfidfVectorizer for text

C MinMaxScaler for numeric, LabelEncoder for categorical, CountVectorizer for text

D Normalizer for numeric, OneHotEncoder for categorical, LabelEncoder for text

💡 Hint

Text data needs vectorization, categorical data needs encoding, numeric data needs scaling.

StandardScaler is common for numeric data. OneHotEncoder is suitable for categorical data to avoid ordinal assumptions. TfidfVectorizer is a good choice for text data to capture word importance.

❓ Hyperparameter

advanced

Effect of 'remainder' parameter in ColumnTransformer

What happens if you set remainder='passthrough' in a ColumnTransformer?

A An error is raised if columns are missing in transformers

B Columns not specified are dropped from the output

C All columns are transformed regardless of specification

D Columns not specified in transformers are passed through without changes

💡 Hint

Think about how to keep columns unchanged in the output.

Setting remainder='passthrough' keeps columns not listed in transformers unchanged and includes them in the output.

🔧 Debug

advanced

Error when fitting ColumnTransformer with incompatible data types

What error will this code raise and why? from sklearn.compose import ColumnTransformer from sklearn.preprocessing import StandardScaler, OneHotEncoder import numpy as np X = np.array([[1, 'red'], [2, 'blue'], [3, 4]]) ct = ColumnTransformer( transformers=[ ('num', StandardScaler(), [0]), ('cat', OneHotEncoder(), [1]) ] ) ct.fit(X)

ML Python

from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import StandardScaler, OneHotEncoder
import numpy as np

X = np.array([[1, 'red'], [2, 'blue'], [3, 4]])

ct = ColumnTransformer(
    transformers=[
        ('num', StandardScaler(), [0]),
        ('cat', OneHotEncoder(), [1])
    ]
)

ct.fit(X)

A TypeError because column 1 has mixed types (strings and integers)

B TypeError because StandardScaler cannot process strings

C No error, the code runs successfully

D IndexError because column 1 does not exist

💡 Hint

Check the data types in the categorical column.

Column 1 has mixed types: strings ('red', 'blue') and an integer (4). OneHotEncoder expects consistent types and raises TypeError.

🧠 Conceptual

expert

Why use ColumnTransformer in a machine learning pipeline?

Which is the best explanation for why ColumnTransformer is useful when working with mixed data types?

A It replaces the need for feature selection by removing irrelevant columns

B It automatically detects data types and chooses the best model

C It allows applying different preprocessing steps to different columns in one step, simplifying pipelines

D It converts all data to numeric format without any user input

💡 Hint

Think about handling numeric and categorical data differently.

ColumnTransformer lets you apply different transformations to different columns in a single step, making preprocessing easier and cleaner.