AI-900 Practice Question: Describe fundamental principles of machine learning on Azure
What is 'dimensionality reduction' and why is it useful in machine learning?
⚠ Common exam trap
It's easy for candidates to confuse dimensionality reduction with model output simplification or hardware reduction, because the word 'reduction' is used broadly, but the exam specifically tests the definition as a feature preprocessing technique for input data.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Reducing the number of input features while preserving key information for efficient modelling
Dimensionality reduction is the process of reducing the number of input features (variables) in a dataset while retaining as much of the original information as possible. This is useful in machine learning because it helps combat the 'curse of dimensionality', reduces overfitting, lowers computational cost, and can improve model performance by eliminating noise and redundant features. In Azure Machine Learning, techniques like Principal Component Analysis (PCA) are commonly used for this purpose.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Reducing the physical size of AI hardware components for edge deployment
Why it's wrong here
Shrinking the physical footprint of AI hardware (e.g., custom ASICs, system-on-chip packaging) is an electronics and mechanical engineering problem that enables edge deployment, but it does not affect the mathematical structure of the dataset. Dimensionality reduction instead operates on the feature matrix, removing redundant or noisy columns while retaining the essential variance needed for modeling. Confusing hardware miniaturisation with feature-space reduction conflates the physical compute platform with the data representation.
- ✓
Reducing the number of input features while preserving key information for efficient modelling
Why this is correct
Dimensionality reduction (e.g., principal component analysis, UMAP, autoencoders) projects a high-dimensional input feature space into a lower-dimensional subspace that retains the majority of the useful signal. This reduces the number of input features while preserving key information, which leads to shorter training times, lower risk of overfitting, and improved interpretability. It is a standard data preprocessing step in Azure Machine Learning pipelines, often used before training classification or regression models.
- ✗
Reducing the model's output to a single dimension for binary decision making
Why it's wrong here
Mapping a model's prediction to a single binary outcome (0 or 1) is a design choice for the output layer of a classifier, often using a sigmoid activation and a threshold. Dimensionality reduction, by contrast, targets the input feature space (the X matrix) and does not constrain the output dimensionality. A binary decision is about the target variable's encoding, not about reducing the number of independent variables.
- ✗
Simplifying the Azure ML workspace to have fewer compute resources and experiments
Why it's wrong here
Configuring an Azure Machine Learning workspace with fewer compute clusters, datastores, or experiment runs is an MLOps governance activity aimed at cost and resource management. That administrative simplification does not alter the dimensionality of the training data. Dimensionality reduction is a purely algorithmic preprocessing technique applied to the raw features before model training, independent of workspace organization.
Go deeper
Related to this question
Learn chapter
Machine Learning Core Concepts
Key term
Model
In IT and AI, a model is a trained mathematical representation that learns patterns from data to make predictions or decisions.
Key term
Overfitting
Overfitting occurs when a machine learning model learns the training data too well, including its noise and outliers, causing it to perform poorly on new, unseen data.
About these practice questions
Courseiva writes every AI-900 question from scratch — 985 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AI-900 practice question is part of Courseiva's free Microsoft certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AI-900 exam.