🕑 8 min read

Introduction

Welcome to Part 3 of the AI-900 Practice Questions series.

The main focus is on machine learning model evaluation, regression, classification, clustering, Azure Machine Learning Designer, feature engineering, computer vision, document processing, and facial recognition.

Each question includes the correct answer and a short explanation for quick revision.

Question 51

Which two metrics can you use to evaluate a regression model?

A. Coefficient of determination (R²)
B. F1 score
C. Root Mean Squared Error (RMSE)
D. Area Under Curve (AUC)
E. Balanced accuracy

Answer: A and C

Explanation: R² and RMSE are regression evaluation metrics, while F1 score and AUC are commonly associated with classification.

Question 52

Predicting how many vehicles will travel across a bridge on a given day is an example of:

A. Classification
B. Clustering
C. Regression
D. Anomaly detection

Answer: C. Regression

Explanation: The model predicts a numerical value—the number of vehicles—so regression is appropriate.

Question 53

You use Azure Machine Learning Designer to build a model that predicts automobile prices.

Which sequence of modules should be used?

A. Select Columns in Dataset → Clean Missing Data → Split Data → Linear Regression → Train Model → Score Model
B. K-Means Clustering → Split Data → Train Model → Score Model
C. Select Columns → Classification → Train Model → Score Model
D. Summarize Data → K-Means → Train Model → Score Model

Answer: A

Explanation: The supplied workflow selects relevant columns, cleans missing values, splits the data, and uses Linear Regression because automobile price is numerical.

Question 54

Which type of machine learning should you use to identify groups of people who have similar purchasing habits?

A. Classification
B. Regression
C. Clustering
D. Anomaly detection

Answer: C. Clustering

Explanation: Clustering groups data points based on similar characteristics without requiring predefined categories.

Question 55

Which type of machine learning model can be used to predict the sale price of auctioned items?

A. Classification
B. Clustering
C. Regression
D. Anomaly detection

Answer: C. Regression

Explanation: Sale price is a numerical value, so a regression model should be used.

Question 56

Which metric can you use to evaluate a classification model?

A. True positive rate
B. Mean Absolute Error (MAE)
C. Coefficient of determination (R²)
D. Root Mean Squared Error (RMSE)

Answer: A. True positive rate

Explanation: True positive rate is a classification metric, while MAE, R², and RMSE are regression metrics.

Question 57

Which two components can you drag onto a canvas in Azure Machine Learning Designer?

A. Dataset
B. Compute
C. Pipeline
D. Module

Answer: A and D

Explanation: According to the supplied material, datasets and modules can be dragged onto the Designer canvas and connected to create machine learning workflows.

Question 58

You need to create training and validation datasets from an existing dataset.

Which Azure Machine Learning Designer module should you use?

A. Select Columns in Dataset
B. Add Rows
C. Split Data
D. Join Data

Answer: C. Split Data

Explanation: Split Data divides an existing dataset into separate subsets for training and evaluation.

Question 59

Match each scenario with the appropriate machine learning type.

Predict how many minutes late a flight will arrive based on snowfall
Segment customers into groups for marketing
Predict whether a student will complete a university course

A. Regression → Clustering → Classification
B. Classification → Regression → Clustering
C. Clustering → Classification → Regression
D. Regression → Classification → Clustering

Answer: A

Explanation: Numerical predictions use regression, grouping similar customers uses clustering, and predicting a yes/no outcome uses classification.

Question 60

What represents the calculated probability of a correct image classification?

A. Accuracy
B. Confidence
C. Root Mean Square Error
D. Sentiment

Answer: B. Confidence

Explanation: The confidence score represents the probability associated with the predicted image class.

Question 61

Ensuring that an AI system does not provide a prediction when important fields contain unusual or missing values demonstrates which responsible AI principle?

A. Inclusiveness
B. Privacy and security
C. Reliability and safety
D. Transparency

Answer: C. Reliability and safety

Explanation: Reliability and safety require AI systems to behave safely when they encounter unexpected or abnormal input.

Question 62

Ensuring that numerical variables in training data are on a similar scale is an example of:

A. Data ingestion
B. Feature engineering
C. Feature selection
D. Model training

Answer: B. Feature engineering

Explanation: The supplied material treats scaling and normalization as feature-engineering activities.

Question 63

Assigning classes to images before training a classification model is an example of:

A. Evaluation
B. Feature engineering
C. Hyperparameter tuning
D. Labeling

Answer: D. Labeling

Explanation: Labeling assigns known categories or target values to training examples before supervised model training.

Question 64

An Azure Machine Learning model predicts product quality using the following fields:

Date
Time
Mass (kg)
Temperature (°C)
Quality Test

Determine whether each statement is correct.

Statement 1: Mass (kg) is a feature.
Statement 2: Quality Test is a label.
Statement 3: Temperature (°C) is a label.

Answers:

Yes
Yes
No

Explanation: Mass and temperature are inputs used to make predictions, while Quality Test is the target label being predicted.

Question 65

Determine whether each machine learning statement is correct.

Statement 1: You train a regression model using unlabeled data.
Statement 2: Classification is used to predict sequential numerical data over time.
Statement 3: Grouping items by common characteristics is an example of clustering.

Answers:

No
No
Yes

Explanation: Regression and classification are supervised techniques, while clustering groups similar items based on their characteristics.

Question 66

Which two actions are performed during the data ingestion and data preparation stage of an Azure Machine Learning process?

A. Calculate model accuracy
B. Score test data using the model
C. Combine multiple datasets
D. Use the model for real-time predictions
E. Remove records that contain missing values

Answer: C and E

Explanation: Combining datasets and handling missing values are data preparation activities performed before model training and evaluation.

Question 67

You need to predict the animal population of an area.

Which type of machine learning should you use?

A. Regression
B. Clustering
C. Classification
D. Anomaly detection

Answer: A. Regression

Explanation: Animal population is a numerical value, so the supplied material identifies regression as the appropriate technique.

Question 68

Which two languages can you use to write custom code for Azure Machine Learning Designer according to the supplied material?

A. Python
B. R
C. C#
D. Scala

Answer: A and B

Explanation: The supplied practice material identifies Python and R as languages for customizing Designer workflows.

Question 69

Determine whether each statement about labels is correct.

Statement 1: For a regression model, labels must be numeric.
Statement 2: For a clustering model, labels must be used.
Statement 3: For a classification model, labels must be numeric.

Answers:

Yes
No
No

Explanation: Regression predicts numeric labels, clustering is unsupervised and does not require labels, and classification labels can represent categories rather than continuous numeric values.

Question 70

Your company wants to build a recycling machine for bottles.

The machine must automatically identify bottles of the correct shape and reject all other items.

Which AI workload should be used?

A. Anomaly detection
B. Conversational AI
C. Computer vision
D. Natural language processing

Answer: C. Computer vision

Explanation: Computer vision analyzes visual information and can identify objects based on their visual characteristics.

Question 71

Determine whether each statement about an object detection model in Custom Vision is correct.

Statement 1: You must choose either Multilabel or Multiclass classification when creating an object detection model.

Statement 2: You can create an object detection model to find the location of content within an image.

Statement 3: You can select from predefined domains when creating an object detection model.

Answers:

No
Yes
Yes

Explanation: According to the supplied material, object detection locates objects in images and supports predefined domains, while Multilabel/Multiclass applies to classification rather than object detection.

Question 72

In which two scenarios can you use Form Recognizer?

A. Extract an invoice number from an invoice
B. Translate a form from French to English
C. Find an image of a product in a catalog
D. Identify the retailer from a receipt

Answer: A and D

Explanation: Form Recognizer can extract structured information from documents such as invoices and receipts.

Question 73

Counting the number of animals in an area based on a video feed is an example of:

A. Forecasting
B. Computer vision
C. Conversational AI
D. Anomaly detection

Answer: B. Computer vision

Explanation: Analyzing objects in images or video streams is a computer vision workload.

Question 74

You have a database containing employees and their photos. You are tagging new photos of the employees.

Determine whether each statement is correct.

Statement 1: The Face service can be used to perform facial recognition for employees.

Statement 2: Providing more sample photos of each employee from different angles can improve recognition.

Statement 3: If an employee is wearing sunglasses, the Face service will always fail to recognize the employee.

Answers:

Yes
Yes
No

Explanation: The supplied material indicates that facial recognition can identify employees, multiple representative images can improve results, and sunglasses do not guarantee recognition failure.

Question 75

Which AI capability is most appropriate for determining the location of objects within an image?

A. Image classification
B. Object detection
C. Regression
D. Natural language processing

Answer: B. Object detection

Explanation: Object detection identifies objects and determines their positions within an image.

Conclusion

That completes AI-900 Practice Questions Part 3 — Questions 51–75.

This section covered several important AI-900 concepts:

Regression → Predict numerical values

Classification → Predict categories or classes

Clustering → Group similar data without predefined labels

Feature Engineering → Transform data into more useful inputs

Model Evaluation → Measure model performance using appropriate metrics

Computer Vision → Analyze images and video

Object Detection → Identify objects and their locations

Document Processing → Extract structured information from documents

A particularly important exam concept is knowing which machine learning technique fits a scenario. If you’re predicting a number, think regression. If you’re predicting a predefined category, think classification. If you’re grouping similar items without known categories, think clustering.

Connect with Me on LinkedIn

Thank you for reading! If you found this article helpful, connect with me on LinkedIn for more learning resources, technology insights, certification content, and career tips. Let’s learn and grow together.