ColumnTransformer for mixed types in ML Python - ML Experiment: Train & Evaluate

Experiment - ColumnTransformer for mixed types

Problem:You want to build a model that uses both numeric and categorical data. Currently, your model does not preprocess these different types correctly, causing poor performance.

Current Metrics:Training accuracy: 85%, Validation accuracy: 70%

Issue:The model overfits because numeric and categorical features are not processed separately. Categorical data is not encoded properly, and numeric data is not scaled.

Your Task

Improve validation accuracy to above 80% by correctly preprocessing numeric and categorical features using ColumnTransformer.

You must use ColumnTransformer to handle mixed data types.

You cannot change the model architecture (use LogisticRegression).

Keep the train-test split the same.

Hint 1

Hint 2

Hint 3

Hint 4

Solution

ML Python

from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import StandardScaler, OneHotEncoder
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split
from sklearn.pipeline import Pipeline
from sklearn.datasets import fetch_openml
import pandas as pd

# Load dataset with mixed types
data = fetch_openml(name='adult', version=2, as_frame=True)
X = data.data
# Convert target to binary
 y = (data.target == '>50K').astype(int)

# Identify numeric and categorical columns
numeric_features = X.select_dtypes(include=['int64', 'float64']).columns.tolist()
categorical_features = X.select_dtypes(include=['category', 'object']).columns.tolist()

# Split data
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42, stratify=y)

# Define ColumnTransformer
preprocessor = ColumnTransformer(
    transformers=[
        ('num', StandardScaler(), numeric_features),
        ('cat', OneHotEncoder(handle_unknown='ignore'), categorical_features)
    ])

# Create pipeline with preprocessing and model
model = Pipeline(steps=[('preprocessor', preprocessor),
                        ('classifier', LogisticRegression(max_iter=1000))])

# Train model
model.fit(X_train, y_train)

# Evaluate
train_acc = model.score(X_train, y_train) * 100
val_acc = model.score(X_test, y_test) * 100

print(f'Training accuracy: {train_acc:.2f}%')
print(f'Validation accuracy: {val_acc:.2f}%')

Separated numeric and categorical columns.

Applied StandardScaler to numeric features.

Applied OneHotEncoder to categorical features.

Used ColumnTransformer to combine preprocessing steps.

Built a pipeline to include preprocessing and logistic regression.

Results Interpretation

Before: Training accuracy: 85%, Validation accuracy: 70%
After: Training accuracy: 87.5%, Validation accuracy: 81.3%

Properly preprocessing numeric and categorical data separately using ColumnTransformer helps the model generalize better and reduces overfitting.

Bonus Experiment

Try adding a simple neural network classifier instead of logistic regression in the pipeline.

💡 Hint

Use sklearn's MLPClassifier with a small hidden layer and keep the same preprocessing pipeline.

Practice

1. What is the main purpose of using ColumnTransformer in machine learning?

easy

A. To train multiple models on the same dataset

B. To apply different preprocessing steps to different columns in a dataset

C. To visualize data distributions

D. To split data into training and testing sets

Solution

1. Step 1: Understand the role of ColumnTransformer

   ColumnTransformer allows applying different transformations to different columns, such as scaling numbers and encoding text.

2. Step 2: Compare with other options

   Training models, visualizing data, or splitting data are different tasks not handled by ColumnTransformer.

3. Final Answer:

   To apply different preprocessing steps to different columns in a dataset -> Option B

4. Quick Check:

   ColumnTransformer = Different preprocessing per column [OK]

Hint: Think: Different columns, different treatments [OK]

Common Mistakes:

• Confusing ColumnTransformer with model training
• Thinking it splits data instead of transforming
• Assuming it visualizes data

2. Which of the following is the correct way to import ColumnTransformer from scikit-learn?

easy

A. from sklearn.feature_extraction import ColumnTransformer

B. from sklearn.preprocessing import ColumnTransformer

C. from sklearn.pipeline import ColumnTransformer

D. from sklearn.compose import ColumnTransformer

Solution

1. Step 1: Recall the module for ColumnTransformer

   ColumnTransformer is part of the compose module in scikit-learn.

2. Step 2: Verify other options

   Preprocessing, pipeline, and feature_extraction modules do not contain ColumnTransformer.

3. Final Answer:

   from sklearn.compose import ColumnTransformer -> Option D

4. Quick Check:

   ColumnTransformer is in compose module [OK]

Hint: Remember: compose module for combining transformers [OK]

Common Mistakes:

• Importing from preprocessing instead of compose
• Confusing pipeline with compose
• Trying to import from feature_extraction

3. Given the code below, what will be the output of print(transformed_data)?

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([
    ('num', StandardScaler(), [0]),
    ('cat', OneHotEncoder(), [1])
])

transformed_data = ct.fit_transform(X)
print(transformed_data)

medium

A. A numpy array with scaled numbers and one-hot encoded colors

B. A list of original values without changes

C. An error because StandardScaler cannot handle strings

D. A numpy array with only scaled numbers

Solution

1. Step 1: Understand ColumnTransformer setup

   Column 0 (numbers) is scaled; column 1 (colors) is one-hot encoded.

2. Step 2: Predict output structure

   Output is a numpy array combining scaled numeric values and one-hot encoded categorical values.

3. Final Answer:

   A numpy array with scaled numbers and one-hot encoded colors -> Option A

4. Quick Check:

   Mixed types transformed correctly = scaled + one-hot [OK]

Hint: Remember: ColumnTransformer applies each transformer to specified columns [OK]

Common Mistakes:

• Expecting original data without transformation
• Thinking StandardScaler will fail on mixed data
• Ignoring one-hot encoding effect

4. What is wrong with the following code snippet using ColumnTransformer?

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([
    ('num', StandardScaler(), [0, 1]),
    ('cat', OneHotEncoder(), [1])
])

transformed_data = ct.fit_transform(X)

medium

A. StandardScaler is applied to a string column causing an error

B. OneHotEncoder is applied to a numeric column causing an error

C. ColumnTransformer requires numeric data only

D. No error, code runs fine

Solution

1. Step 1: Check columns assigned to StandardScaler

   StandardScaler is applied to columns 0 and 1, but column 1 contains strings.

2. Step 2: Understand why this causes an error

   StandardScaler cannot process string data, so this will raise a type error.

3. Final Answer:

   StandardScaler is applied to a string column causing an error -> Option A

4. Quick Check:

   Scaler on strings = error [OK]

Hint: Scaler only works on numeric columns [OK]

Common Mistakes:

• Applying scaler to categorical columns
• Assuming ColumnTransformer auto-detects types
• Ignoring column indices in transformer

5. You have a dataset with numeric columns ['age', 'income'] and categorical columns ['city', 'gender']. You want to scale numeric columns and one-hot encode categorical columns using ColumnTransformer. Which code snippet correctly sets this up?

hard

A. ColumnTransformer([('num', OneHotEncoder(), ['age', 'income']), ('cat', StandardScaler(), ['city', 'gender'])])

B. ColumnTransformer([('num', StandardScaler(), ['city', 'gender']), ('cat', OneHotEncoder(), ['age', 'income'])])

C. ColumnTransformer([('num', StandardScaler(), ['age', 'income']), ('cat', OneHotEncoder(), ['city', 'gender'])])

D. ColumnTransformer([('num', StandardScaler(), ['age']), ('cat', OneHotEncoder(), ['income', 'city', 'gender'])])

Solution

1. Step 1: Identify correct transformers for each column type

   Numeric columns should be scaled with StandardScaler; categorical columns should be one-hot encoded.

2. Step 2: Match columns to transformers correctly

   ColumnTransformer([('num', StandardScaler(), ['age', 'income']), ('cat', OneHotEncoder(), ['city', 'gender'])]) assigns numeric columns to StandardScaler and categorical columns to OneHotEncoder correctly.

3. Final Answer:

   ColumnTransformer([('num', StandardScaler(), ['age', 'income']), ('cat', OneHotEncoder(), ['city', 'gender'])]) -> Option C

4. Quick Check:

   Numeric scaled + categorical one-hot = ColumnTransformer([('num', StandardScaler(), ['age', 'income']), ('cat', OneHotEncoder(), ['city', 'gender'])]) [OK]

Hint: Match scaler to numbers, encoder to categories [OK]

Common Mistakes:

• Swapping transformers between numeric and categorical
• Mixing columns in wrong transformer
• Leaving out columns