Feature engineering pipelines in MLOps - Mini Project: Build & Apply

Feature Engineering Pipelines

📖 Scenario: You are working on a machine learning project. You need to prepare your data by creating a feature engineering pipeline. This pipeline will help you clean and transform your data automatically before training your model.

🎯 Goal: Build a simple feature engineering pipeline using scikit-learn that scales numeric features and encodes categorical features.

📋 What You'll Learn

Create a dataset dictionary with numeric and categorical features

Define a configuration variable for numeric feature names

Build a pipeline that scales numeric features and encodes categorical features

Print the transformed feature array

💡 Why This Matters

🌍 Real World

Feature engineering pipelines automate data preparation steps, making machine learning workflows faster and less error-prone.

💼 Career

Understanding how to build and use feature engineering pipelines is essential for MLOps engineers and data scientists to deploy reliable ML models.

1

Create the initial dataset dictionary

Create a dictionary called data with these exact entries: 'age': [25, 32, 47], 'salary': [50000, 60000, 80000], and 'department': ['sales', 'engineering', 'marketing'].

MLOps

# Your code here

Need a hint?

Use a dictionary with keys 'age', 'salary', and 'department' and assign lists of values exactly as shown.

2

Define numeric feature names

Create a list called numeric_features containing the strings 'age' and 'salary'.

MLOps

data = {
    'age': [25, 32, 47],
    'salary': [50000, 60000, 80000],
    'department': ['sales', 'engineering', 'marketing']
}
# Your code here

Need a hint?

Define a list named numeric_features with the exact two strings.

3

Build the feature engineering pipeline

Import ColumnTransformer, StandardScaler, and OneHotEncoder from sklearn. Create a ColumnTransformer called preprocessor that applies StandardScaler() to numeric_features and OneHotEncoder() to the 'department' feature.

MLOps

data = {
    'age': [25, 32, 47],
    'salary': [50000, 60000, 80000],
    'department': ['sales', 'engineering', 'marketing']
}
numeric_features = ['age', 'salary']
# Your code here

Need a hint?

Use ColumnTransformer with two transformers: one for numeric features using StandardScaler() and one for categorical feature 'department' using OneHotEncoder().

4

Transform and print the processed features

Convert data to a pandas DataFrame called df. Use preprocessor.fit_transform(df) to transform the data and assign it to features. Print features.

MLOps

from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import StandardScaler, OneHotEncoder
import pandas as pd

data = {
    'age': [25, 32, 47],
    'salary': [50000, 60000, 80000],
    'department': ['sales', 'engineering', 'marketing']
}
numeric_features = ['age', 'salary']

preprocessor = ColumnTransformer(
    transformers=[
        ('num', StandardScaler(), numeric_features),
        ('cat', OneHotEncoder(), ['department'])
    ]
)
# Your code here

Need a hint?

Use pandas.DataFrame to convert data. Then call preprocessor.fit_transform(df) and print the result.