Magnimind Machine Learning Quiz

Welcome to your Magnimind Machine Learning Quiz

Name

Email

1.

In L1 regularization, we penalize the absolute value of the weights while in L2 regularization, we penalize the squared value of the weights.

True

False

2.

The scikit-learn Python machine learning library provides the ColumnTransformer that allows you to selectively apply data transforms to different columns in your dataset

True

False

3.

It is good practice to use MinMaxScaling on a feature with a few extreme outliers.

True

False

4.

The parameter k should take odd values in kNN, so that there are no ties in the voting.

True

False

5.

Treating a non-ordinal categorical variable as continuous variable would result in a better predictive model.

True

False

6.

One-hot-encoding increasing the dimensionality of a data set.

True

False

7.

OLS Regression is expected to have more overfitting (lower bias) than Ridge.

True

False

8.

Fitting your scaling transformation separately to your training and the test sets improves the model performance.

True

False

9.

The more features that we use to represent our data, the better the learning algorithm will generalize to new data points.

True

False

10.

It is not a good machine learning practice to use the test set to help adjust the hyperparameters of your learning algorithm

True

False

11.

Cross validation:

Does nothing to prevent overfitting

Is often used to select hyperparameters

Is guaranteed to prevent overfitting

None of the above

12.

Say, there are two kids Jack and Jill in a maths exam. Jack only learnt additions and Jill memorized the questions and their answers from the maths book. Now, who will succeed in the exam? The answer is neither. From machine learning lingo, Jack is blank1 and Jill is blank2.

blank1=overfitting, blank2=underfitting

blank1=underfitting, blank2=overfitting

blank1=underfitting, blank2=underfitting

blank1=overfitting, blank2=overfitting

13.

You are working on a classification problem. For validation purposes, you have randomly sampled the training data set into train and validation. You are confident that your model will work incredibly well on unseen data since your validation accuracy is high. However, you get shocked after getting poor test accuracy. What might have gone wrong?

You trained an OLS model.

You are using an imbalanced dataset.

You used the simplest model you could come up with

You used random sampling instead of stratified sampling when forming the test and train datasets.

14.

Which of the following models is more acceptable then others to be implemented on new data points?

My model achieves a training error lower than all previous models!

My model achieves a test error lower than all previous models! (When regularization parameter α is chosen so as to minimize test error.)

My method achieves a cross-validation error lower than all previous methods! (When regularization parameter α is chosen so as to minimize crossvalidaton error.)

My method achieves a test error lower than all previous methods! (When regularization parameter α is chosen so as to minimize crossvalidaton error.)

15.

Which of the following is not correct regarding LogisticRegression and LinearSVM?

Both algorithms generally perform comparably in practice.

LogisticRegression uses hinge-loss function while LinearSVM uses sigmoid function.

Different from Logistic Regression using α as the parameter in front of regularized term to control the weight of regularization, correspondingly, SVM uses C in front of fit term.

LogisticRegression provides probability values for the predictions.