Intermediate ML Fundamentals Quiz

1. What is the relationship between R² and the coefficient of determination?

R² measures the proportion of variance explained by the model

R² and coefficient of determination are unrelated

R² only applies to classification models

R² cannot be negative

2. What does overfitting indicate about a model's performance?

The model performs well on both training and test data

The model learns training noise and performs poorly on unseen data

The model has too few parameters

The model is too simple to capture patterns

3. In support vector machines, what is the kernel trick used for?

To speed up training on large datasets

To map data into higher-dimensional space without explicit computation

To handle missing values automatically

To reduce the number of support vectors

4. How does stratified k-fold cross-validation differ from standard k-fold?

It uses fewer folds for faster computation

It maintains class distribution within each fold

It only applies to regression problems

It eliminates the need for a test set

5. What is the primary goal of feature engineering in machine learning?

To increase the dataset size

To create meaningful features that improve model performance

To eliminate all categorical variables

To reduce computational time only

6. In gradient descent, what does the learning rate control?

The number of iterations required to converge

The size of steps taken toward the minimum

The direction of weight updates

The batch size used in training

7. Which of the following is a valid approach to handle missing data?

Delete rows with any missing values

Imputation using mean, median, or forward-fill methods

Ignore missing values during model training

Convert all missing values to zero

8. What is the main difference between supervised and unsupervised learning?

Supervised learning requires labeled data; unsupervised does not

Unsupervised learning is always more accurate

Supervised learning cannot handle large datasets

Unsupervised learning requires a validation set

9. In neural networks, what is the role of the activation function?

To initialize network weights randomly

To introduce non-linearity and enable learning of complex patterns

To reduce the number of parameters

To normalize input features

10. Which technique is used to handle multicollinearity in regression?

Increasing the number of features

Ridge regression or removing correlated features

Using only categorical variables

Increasing the learning rate

11. Which of the following best describes the bias-variance tradeoff in machine learning?

High bias leads to underfitting; high variance leads to overfitting

High bias and high variance always result in poor model performance

Bias and variance are unrelated to model accuracy

Variance measures the model's ability to generalize to new data

12. In ensemble methods like Random Forest, what advantage does bagging provide?

Reduces variance by averaging predictions from multiple models

Eliminates the need for hyperparameter tuning

Ensures every feature is used equally

Guarantees perfect accuracy on test data

13. What is the purpose of a confusion matrix in classification?

To display feature importance rankings

To summarize true positives, false positives, true negatives, and false negatives

To visualize the relationship between two features

To show the distribution of target classes

14. Which of the following is a disadvantage of decision trees?

They cannot handle non-linear relationships

They are prone to overfitting on training data

They require feature scaling

They cannot work with categorical variables

15. In k-means clustering, what determines the final cluster assignments?

The initial random seed only

Distance to the nearest cluster centroid

The order of data points in the dataset

The number of features in the data

16. What is the main objective of Principal Component Analysis (PCA)?

To classify data into predefined categories

To reduce dimensionality while preserving variance

To handle missing values in datasets

To balance imbalanced datasets

17. Feature scaling is important in which of the following algorithms?

Decision Trees only

KNN and SVM

Linear Regression only

Random Forests only

18. Which metric is most appropriate for evaluating a highly imbalanced classification problem?

Accuracy

Precision and Recall (F1-score)

Mean Squared Error

Root Mean Squared Error

19. What does regularization do in machine learning models?

Increases model complexity to fit training data better

Penalizes large coefficients to prevent overfitting

Removes outliers from the dataset

Improves computational efficiency

20. In cross-validation, what is the primary purpose of splitting data into k folds?

To increase the size of the training set

To provide a more robust estimate of model performance

To reduce computational cost

To eliminate the need for a test set

General Data Science Intermediate ML Fundamentals Quiz