Quiz – Split Data Engineering

Question 1 of 6

1. Question

Feature creation

Information leakage between training, validation, and testing

Correct Incorrect Correct answer

Any model training

Correct Incorrect Correct answer

Data collection

Correct Incorrect Correct answer

Feature creation

Correct Incorrect Correct answer

Question 2 of 6

2. Question

Storing raw data

Detecting overfitting and guiding hyperparameter selection

Correct Incorrect Correct answer

Optimizing the final weights only

Correct Incorrect Correct answer

Reporting the final unbiased score

Correct Incorrect Correct answer

Storing raw data

Correct Incorrect Correct answer

Question 3 of 6

3. Question

Data cleaning

An unbiased final performance estimate

Correct Incorrect Correct answer

Parameter optimization

Correct Incorrect Correct answer

Hyperparameter tuning

Correct Incorrect Correct answer

Data cleaning

Correct Incorrect Correct answer

Question 4 of 6

4. Question

Prevents all overfitting on its own

Optimizes the model parameters

Correct Incorrect Correct answer

Gives the final generalization score

Correct Incorrect Correct answer

Replaces the validation set

Correct Incorrect Correct answer

Prevents all overfitting on its own

Correct Incorrect Correct answer

Question 5 of 6

5. Question

It has no real purpose

To keep the final estimate unbiased and reflect true generalization

Correct Incorrect Correct answer

Because it is used to tune hyperparameters

Correct Incorrect Correct answer

So it can help train the model

Correct Incorrect Correct answer

It has no real purpose

Correct Incorrect Correct answer

Question 6 of 6

6. Question

Match each data split to its role.

Sort elements

Optimize model parameters
Tune hyperparameters and detect overfitting
Give an unbiased final estimate
Prevent information leakage

Training set

Correct Incorrect Correct answer

Validation set

Correct Incorrect Correct answer

Test set

Correct Incorrect Correct answer

Partitioning

Correct Incorrect Correct answer