STINTSY Quiz 2

In evaluating the performance of a linear regression model, which metric is more commonly used when you want to minimize the impact of large errors and also retain the original scale of the target variable?

• Mean Sum of Squared Error

• Coefficient of Determination

• Root Mean Squared Error

• Mean of Absolute Error

In the context of supervised learning with linear regression, what is the primary objective when fitting a model to the training data?

• Maximize the number of features in the model

• Minimize the differences between predicted and actual target values

• Ensure that the model perfectly fits the training data

• Maximize the R-squared value regardless of other metrics

In the context of training a linear regression model using gradient descent, what is the potential consequence of choosing a learning rate that is too high?

• The model converges too slowly

• The model converges to the optimal solution faster

• The model performs better on unseen data

• The model may fail to converge and oscillate around the minimum

What is a key difference between stochastic gradient descent (SGD) and mini-batch gradient descent in the context of training a linear regression model?

• SGD uses all training data at once, while mini-batch only uses a single sample

• SGD updates weights after each data point, while mini-batch updates weights after a small subset of data

• Mini-batch gradient descent is faster but less accurate than SGD

• Mini-batch gradient descent always converges to the global minimum, while SGD does not

Which of the following feature preprocessing steps is typically important before applying linear regression?

• Scaling or normalizing numerical features

• Adding polynomial features to increase dimensionality

• Removing outliers without any consideration

• Encoding continuous features as categorical