Building, Training, and Validating Models in Microsoft Azure

Question 1

It is required that a predictions are made weekly to find out which customers are likely default on their loan repayment? A designed model should be deployed to?

Real-time inference end-point

Batch inference end-point

Pipeline

Web service

Answer 1

Batch inference end-point

Question 2

Calvis Hammond wants to track how a new model will perform upon deployment. Which is the best to do?

Perform AB testing of different models

Deploy new model

Centralize data pipeline

Answer 2

Perform AB testing of different models

Question 3

A _____ variable changes in response to the _______ variable

Dependent, independent

Independent, dependent

Indepentent, real

Dependent, propriatory

Answer 3

Dependent, independent

Question 4

We run into modelling difficulties when the

Number of sample space < number of predictors

Number of sample space = number of predictors

Number of sample space > number of predictors

Answer 4

Number of sample space < number of predictors

Question 5

Participants rated music playing at a fast tempo and at a slow tempo the same in terms of their happiness. What type of hypothesis is this?

Null hypothesis

Alternative hypothesis

Derived hypothesis

Research hypothesis

Answer 5

Null hypothesis

Question 6

A doctor predicts that a patient is sick, when they are actually in good health. What type of error did the doctor make?

Type II error

This is not an error

Type I

Answer 6

Type I

Question 7

_________ are not distance-based models and can handle varying ranges of features. Hence, Scaling is not required while modelling trees.

Tree based models

Principal Component Analysis(PCA

k-nearest neighbors

Answer 7

Tree based models

Question 8

How should you proceed with splitting data when there is an important temporal aspect to the data?

Stratified sampling

Non-random sampling

Random sampling

Answer 8

Non-random sampling

Question 9

There are several specific types of supervised learning that are represented within Azure Machine Learning Studio (classic): classification, regression, and ___________

Anomaly detection

Clustering

Recommendation

Answer 9

Anomaly detection

“After setting model parameters, you must train the model by using a labeled data set and the Train Anomaly Detection Model training module. “

Question 10

A model deployed by a data scientist starts experiencing performance degradation. What is the possible cause of this?

Lack of monitoring

Inability to validate

Model drift

Answer 10

Model drift