Data Science Data Wrangling and Preprocessing 5

Question 1

When concatenating training feature engineering, fitting an encoder on the full dataset before train/test split causes what problem?

Accepted Answer

Data leakage

Answer

Underfitting

Answer

Class imbalance

Answer

Vanishing gradients

Question 2

A target variable has 95% class A and 5% class B. Which preprocessing technique addresses this imbalance?

Accepted Answer

Resampling such as SMOTE or undersampling

Answer

Min-max scaling

Answer

Datetime parsing

Answer

Log transform

Question 3

Why should imputation of missing values inside a cross-validation loop be done within each fold rather than once beforehand?

Accepted Answer

To avoid leaking information across folds

Answer

To save memory

Answer

To increase the dataset size

Answer

To force normalization

Question 4

Two highly correlated numeric features (r=0.98) are kept in a linear model. What issue does this raise?

Accepted Answer

Multicollinearity

Answer

Missingness

Answer

Skewness

Answer

Sentinel encoding

Question 5

A free-text 'notes' column needs to become model features. Which preprocessing step is most appropriate first?

Accepted Answer

Text vectorization (e.g., TF-IDF or tokenization)

Answer

Min-max scaling

Answer

Median imputation

Answer

Duplicate removal

Question 6

When using scikit-learn, bundling scaling, encoding, and the model into a single object that fits in order prevents leakage and simplifies deployment. What is it called?

Accepted Answer

A Pipeline

Answer

A DataFrame

Answer

A GridSearch

Answer

A ColumnIndex

Question 7

You scale numeric columns but leave one-hot encoded binary columns unscaled. Why is this generally acceptable?

Accepted Answer

Binary 0/1 columns are already on a comparable bounded scale

Answer

Scaling deletes binary columns

Answer

Binary columns cannot be scaled

Answer

It increases dimensionality