Creating interaction features in Data Analysis Python - Practice Exercises

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of interaction feature multiplication

What is the output DataFrame after creating an interaction feature by multiplying columns 'A' and 'B'?

Data Analysis Python

import pandas as pd

df = pd.DataFrame({'A': [1, 2, 3], 'B': [4, 5, 6]})
df['A_B'] = df['A'] * df['B']
print(df)

A {'A': [1, 2, 3], 'B': [4, 5, 6], 'A_B': [4, 5, 6]}

B {'A': [1, 2, 3], 'B': [4, 5, 6], 'A_B': [5, 7, 9]}

C {'A': [1, 2, 3], 'B': [4, 5, 6], 'A_B': [1, 2, 3]}

D {'A': [1, 2, 3], 'B': [4, 5, 6], 'A_B': [4, 10, 18]}

💡 Hint

Multiply each value in column 'A' by the corresponding value in column 'B'.

The interaction feature 'A_B' is created by multiplying each element of 'A' with the corresponding element of 'B'. So 1*4=4, 2*5=10, 3*6=18.

❓ data_output

intermediate

Number of unique interaction features created

Given a DataFrame with columns 'X', 'Y', and 'Z', how many unique pairwise interaction features can be created by multiplying two different columns?

Data Analysis Python

import pandas as pd

df = pd.DataFrame({'X': [1,2], 'Y': [3,4], 'Z': [5,6]})

# Interaction features are created by multiplying pairs of different columns
pairs = [(a,b) for i,a in enumerate(df.columns) for b in df.columns[i+1:]]
print(len(pairs))

A 3

B 6

C 9

D 2

💡 Hint

Count all unique pairs without repetition or order.

With 3 columns, the number of unique pairs is 3 choose 2 = 3. The pairs are (X,Y), (X,Z), and (Y,Z).

🔧 Debug

advanced

Identify the error in interaction feature creation

What error will this code raise when trying to create an interaction feature by multiplying columns 'A' and 'B'?

Data Analysis Python

import pandas as pd

df = pd.DataFrame({'A': [1, 2, 3], 'B': ['4', '5', '6']})
df['A_B'] = df['A'] * df['B']

A ValueError: invalid literal for int() with base 10

B TypeError: unsupported operand type(s) for *: 'int' and 'str'

C No error, output is a column of strings repeated

D TypeError: can't multiply sequence by non-int of type 'int'

💡 Hint

Check the data types of columns before multiplication.

Column 'B' contains strings, so multiplying an int by a str raises a TypeError about unsupported operand types.

🚀 Application

advanced

Creating polynomial interaction features with scikit-learn

Which code snippet correctly creates interaction features (degree 2) from a numeric DataFrame using scikit-learn's PolynomialFeatures?

Data Analysis Python

import pandas as pd
from sklearn.preprocessing import PolynomialFeatures

df = pd.DataFrame({'X': [1, 2], 'Y': [3, 4]})
poly = PolynomialFeatures(degree=2, interaction_only=True, include_bias=False)
X_poly = poly.fit_transform(df)
print(X_poly)

A

poly = PolynomialFeatures(degree=2, interaction_only=False, include_bias=True)
X_poly = poly.fit_transform(df)

B

poly = PolynomialFeatures(degree=3, interaction_only=True, include_bias=False)
X_poly = poly.fit_transform(df)

C

poly = PolynomialFeatures(degree=2, interaction_only=True, include_bias=False)
X_poly = poly.fit_transform(df)

D

poly = PolynomialFeatures(degree=2, interaction_only=True, include_bias=True)
X_poly = poly.fit_transform(df)

💡 Hint

Interaction features exclude powers of single features and bias adds a column of ones.

Setting interaction_only=True creates only interaction terms without powers. include_bias=False excludes the constant term. Degree=2 creates pairwise interactions.

🧠 Conceptual

expert

Effect of interaction features on model complexity

How does adding interaction features between numeric variables affect the complexity of a linear regression model?

A It increases the number of features, which can improve model flexibility but may cause overfitting if not controlled.

B It decreases the number of features, simplifying the model and reducing overfitting risk.

C It has no effect on model complexity because interaction features are ignored by linear regression.

D It always improves model accuracy without any risk of overfitting.

💡 Hint

Think about how adding new features changes the model's ability to fit data.

Adding interaction features increases the number of predictors, allowing the model to capture more complex relationships but also increasing the risk of fitting noise.