Quiz 13

As we increase the number of features in a model, both the bias and variance of

our predictions should decrease.

• True

• False

Stochastic Gradient Descent (SGD) needs to go through the individual data

samples to update the model parameters.

• True

• False

While training your regression model using Gradient Decent, you noticed that the

value of loss increases with epochs, this behavior can be due to high learning

rate.

• True

• False

If you would run 5-fold cross validation on a dataset of 100 examples, you will

need 20 examples for the training set in each run.

• True

• False

If you would run 5-fold cross validation on a dataset of 1000 examples, you will

need (a)   examples for the validation set in each run.

Mention reasons for overfitting ??

• Increase the model complexity (order of the hypothesis function)

• Increase number of used features

• Use small training dataset

• all of them

• None of them

Ridge regression ??

• L1

• L2

Lasso Regression ??

• L1

• L2

... is unsupervised learning algorithm.

• Clustering

• Linear Regression

• Classification

The ... problem occurs when the model performs well in the training set, but badly in the test set.

• Overfitting

• Underfitting

• Fitting