31: COOSE THE COORECT ANSWER :

Q31 : Imagine you are solving a classification problem with a strong class imbalance. One of the classes (prevailing) is 99% of all values in the sample.

Your model has an accuracy of 99% after predictions based on test data. Which of the following is true then?

1. Metric accuracy - not suitable for problems with class imbalance.

2. Metric accuracy - suitable for problems with class imbalance.

3. The precision and recall metrics are suitable for problems with class imbalance.

4. The precision and recall metrics are not suitable for problems with class imbalance.

A- Only 1 and 3

B- Only 1 and 4

C- Only 2 and 3

D- Only 3 and 4

____________________________

Q32 : There are nonlinear data and discontinuities in the input that you suspect might affect the output of your model. What analytical method would you suggest?

1- Decision Trees

2- ANOVA

3- K-means Clustering

4- Linear Regression

____________________________

Q33 : How to choose an attribute to split in a decision tree algorithm?

1- You choose the attribute with the highest infoGain

2- You select the attribute with the smallest infoGain

3- You choose the attribute where the conditional entropy is maximum

4- You select an attribute where the conditional entropy is higher than the base entropy