Where This Fits in the Exam

• Exam Domain: Describe fundamental principles of machine learning on Azure (15–20%)
• Sub-Domain: Identify common machine learning techniques
• Topic: Identify classification machine learning scenarios

On the AI-900 exam, classification questions test your ability to recognize when classification is the appropriate machine learning technique, not how to build models.

---

What Is Classification in Machine Learning?

Classification is a type of supervised machine learning used to predict a category, class, or label.

• The model is trained on labeled data
• The output is discrete, not numeric
• The goal is to decide which category something belongs to

Key exam rule:
If the output is a label or category, the scenario is classification.

---

Characteristics of Classification Scenarios

A classification workload typically includes:

• Historical data with known labels
• Input features used to make predictions
• A finite set of possible outcomes
• Binary or multi-class results

Common classification outputs:

• Yes / No
• True / False
• Approved / Rejected
• Spam / Not Spam
• High Risk / Low Risk

---

Binary vs Multi-Class Classification

Binary Classification

• Only two possible outcomes
• Examples:
  • Fraud / Not Fraud
  • Pass / Fail
  • Churn / No Churn

Multi-Class Classification

• More than two categories
• Examples:
  • Product category (electronics, clothing, food)
  • Support ticket priority (low, medium, high)
  • Image labels (cat, dog, bird)

Both are classification scenarios on the AI-900 exam.

---

Common Classification Use Cases

Decision-Based Predictions

• Loan approval decisions
• Insurance claim approval
• Credit risk classification

Detection and Filtering

• Spam email detection
• Fraud detection
• Content moderation

Categorization

• Customer churn prediction
• Sentiment categories (positive, neutral, negative)
• Product classification

All of these involve choosing a label, not predicting a number.

---

Classification vs Other ML Techniques

Understanding how classification differs from regression and clustering is critical for AI-900.

Technique	Output	Example
Regression	Numeric value	Predicting house price
Classification	Category or label	Approving a loan
Clustering	Group assignment	Customer segmentation

Exam tip:
If the answer choices include Yes/No, True/False, or named groups, think Classification.

---

Example Exam Scenarios

Scenario 1

A bank wants to determine whether a transaction is fraudulent.

• Output: Fraud / Not Fraud
• ML Technique: Classification

---

Scenario 2

A company wants to predict whether a customer will cancel their subscription.

• Output: Cancel / Not Cancel
• ML Technique: Classification

---

Scenario 3

An AI system categorizes customer support tickets into predefined issue types.

• Output: Issue category
• ML Technique: Classification

---

Azure Context for AI-900

On the AI-900 exam, classification scenarios are often described using Azure Machine Learning concepts such as:

• Training models with labeled datasets
• Predicting predefined categories
• Evaluating model accuracy

You are not required to:

• Select algorithms
• Write code
• Configure Azure services

Focus on recognizing the technique, not implementing it.

---

Common Exam Traps and Misconceptions

• ❌ Predicting a numeric score → Regression
• ❌ Grouping data without labels → Clustering
• ❌ Predicting ranges like High / Medium / Low → Classification, not regression
• ✅ Predicting labels or categories → Classification

---

Key Takeaways for the Exam

• Classification predicts categories or labels
• It is a supervised learning technique
• Outputs are discrete, not numeric
• Binary and multi-class scenarios are both classification
• Look for keywords like classify, detect, assign, categorize

---

Go to the Practice Exam Questions for this topic.

Go to the AI-900 Exam Prep Hub main page.