Why engineered features improve models in ML Python - Challenge Your Understanding

Challenge - 5 Problems

🧠 Conceptual

intermediate

Why do engineered features help machine learning models?

Imagine you want to predict house prices. You have raw data like size in square feet and number of rooms. Why might creating new features like 'price per room' help the model?

A Because models only work with features created by humans, not raw data.

B Because new features can reveal hidden patterns that raw data alone might not show.

C Because raw data is always noisy and engineered features remove all noise.

D Because adding more features always makes the model more complex and accurate.

💡 Hint

Think about how combining simple data points can create more meaningful information.

Engineered features combine or transform raw data to highlight important relationships. This helps models learn better patterns and improve predictions.

❓ Predict Output

intermediate

Output of feature scaling on data

What is the output of the following Python code that scales a feature using min-max scaling?

ML Python

import numpy as np
from sklearn.preprocessing import MinMaxScaler

X = np.array([[10], [20], [30], [40], [50]])
scaler = MinMaxScaler()
X_scaled = scaler.fit_transform(X)
print(X_scaled.flatten())

A [0.0 0.25 0.5 0.75 1.0]

B [1.0 0.75 0.5 0.25 0.0]

C [10 20 30 40 50]

D [0.0 0.2 0.4 0.6 0.8]

💡 Hint

Min-max scaling transforms values to a range between 0 and 1 based on min and max values.

The smallest value (10) becomes 0, the largest (50) becomes 1, and others are scaled linearly between.

❓ Model Choice

advanced

Choosing a model for engineered polynomial features

You created polynomial features (squares and cubes) from your original data. Which model below is best suited to use these features effectively?

A Linear regression with regularization (like Ridge or Lasso)

B Decision tree classifier

C Linear regression without regularization

D K-means clustering

💡 Hint

Think about how adding many polynomial features can cause overfitting and how regularization helps.

Regularized linear regression controls complexity from many polynomial features, improving generalization.

❓ Metrics

advanced

Effect of engineered features on model accuracy

A model trained on raw features has 75% accuracy. After adding engineered features, accuracy rises to 85%. What does this improvement most likely indicate?

A The training data was too small to learn from raw features.

B The model is overfitting and will perform worse on new data.

C The model's hyperparameters were changed to increase accuracy.

D Engineered features helped the model capture more useful information.

💡 Hint

Think about what adding meaningful features does to model learning.

Better accuracy usually means the model learned better patterns thanks to the new features.

🔧 Debug

expert

Why does adding engineered features sometimes hurt model performance?

Consider a model where adding many engineered features caused test accuracy to drop. Which reason below best explains this?

A Engineered features always reduce model performance if not normalized.

B The model cannot handle more than 10 features due to algorithm limits.

C The new features introduced noise or irrelevant information causing overfitting.

D The training data was too large, confusing the model.

💡 Hint

Think about how adding many features can sometimes confuse the model.

Too many irrelevant or noisy features can cause the model to fit training data too closely and perform worse on new data.