Explore: All GATE DA topics · GATE DA Data Science and Artificial Intelligence PYQs · GATE DA hub · Syllabus & pattern · GATE DA 2026
How often Machine Learning is asked
In the 3 official GATE DA papers we tagged (2024–2026), Machine Learning (Data Science and Artificial Intelligence) had 27 questions, appearing in 3 of them. Asked in 3 of 3 years. Weightage: 14 marks a year on average (13–15), about 14% of a 100-mark paper. The concepts asked in the most papers: Neural Networks, PCA and Dimensionality Reduction, Hierarchical Clustering.
Year-wise question count
Asked in 3 of 3 years.
| Year | Questions | Marks |
|---|---|---|
| GATE DA 2026 | 8 | 13 |
| GATE DA 2025 | 9 | 14 |
| GATE DA 2024 | 10 | 15 |
Most-asked concepts
Sorted by the number of papers a concept appeared in.
| Concept | Questions | Papers | Years |
|---|---|---|---|
| Neural Networks | 4 | 3 | 2026, 2025, 2024 |
| Hierarchical Clustering | 3 | 3 | 2026, 2025, 2024 |
| PCA and Dimensionality Reduction | 3 | 3 | 2026, 2025, 2024 |
| Support Vector Machines | 3 | 2 | 2025, 2024 |
| Linear Discriminant Analysis | 2 | 2 | 2025, 2024 |
| Naive Bayes | 2 | 2 | 2025, 2024 |
| Model Evaluation | 2 | 1 | 2026 |
| Regularization | 2 | 1 | 2026 |
| Clustering (k-Means) | 1 | 1 | 2024 |
| Cross-Validation | 1 | 1 | 2026 |
| Decision Trees | 1 | 1 | 2024 |
| k-Nearest Neighbours | 1 | 1 | 2024 |
| Linear Regression | 1 | 1 | 2025 |
| Perceptron | 1 | 1 | 2025 |
How to practise Machine Learning PYQs
- Start with the concepts at the top of the table. They have come back the most.
- Solve one year’s Machine Learning questions from the official PDF against the clock, then check the answer listed below.
- For each mistake, note whether you didn’t know the concept, slipped on a calculation or misread the question. Each needs a different fix.
- Once a concept is clean in two papers, move to the next. Then sit a full paper from the year pages.
Questions by year
Each line gives the question number, type, concept, our one-line summary and the official answer, with a link to the official PDF.
GATE DA 2026
- Q.11 · MCQ · 1 mark · PCA and Dimensionality Reduction: Angle between the first and tenth principal components (official answer: B) PDF
- Q.12 · MCQ · 1 mark · Cross-Validation: Number of validation splits produced by leave-one-out cross-validation (official answer: C) PDF
- Q.23 · MSQ · 1 mark · Model Evaluation: Match machine learning tasks with suitable algorithms (official answer: B) PDF
- Q.36 · MCQ · 2 marks · Hierarchical Clustering: First pair merged in agglomerative clustering with Manhattan distance (official answer: D) PDF
- Q.37 · MCQ · 2 marks · Regularization: True statement about the ridge regression regularizer (official answer: D) PDF
- Q.47 · MSQ · 2 marks · Model Evaluation: Accuracy, precision and recall from author classification counts (official answer: A, B) PDF
- Q.55 · NAT · 2 marks · Regularization: Regularised loss for one ridge regression sample under mean absolute error (official answer: 6.90–7.10) PDF
- Q.56 · NAT · 2 marks · Neural Networks: Number of weights in a bias-free multi-layer perceptron (official answer: 135) PDF
GATE DA 2025
- Q.22 · MCQ · 1 mark · Perceptron: Effect of a perceptron update on the score of a misclassified point (official answer: A) PDF
- Q.30 · MSQ · 1 mark · Hierarchical Clustering: Max and min cluster dissimilarities used by single and complete linkage (official answer: B, D) PDF
- Q.34 · NAT · 1 mark · Linear Regression: Least-squares slope of a no-intercept line through three points (official answer: 0.285–0.287) PDF
- Q.35 · NAT · 1 mark · Naive Bayes: Misclassification probability of a Bayes classifier with given priors and likelihoods (official answer: 0.39–0.41) PDF
- Q.42 · MCQ · 2 marks · Neural Networks: Partial derivatives of a small ReLU network output with respect to weights (official answer: A) PDF
- Q.48 · MSQ · 2 marks · Neural Networks: Continuity and differentiability properties of the ReLU activation (official answer: A, C) PDF
- Q.53 · MSQ · 2 marks · Support Vector Machines: Weights, support vectors and margin of a hard-margin SVM on four points (official answer: B, C) PDF
- Q.55 · MSQ · 2 marks · Linear Discriminant Analysis: Properties of a nearest class mean classifier decision function (official answer: B, C) PDF
- Q.60 · NAT · 2 marks · PCA and Dimensionality Reduction: Variance of data projected on the top principal direction (official answer: 100) PDF
GATE DA 2024
- Q.17 · MCQ · 1 mark · Support Vector Machines: Possible support vectors of a hard-margin linear SVM on six points (official answer: D) PDF
- Q.18 · MCQ · 1 mark · PCA and Dimensionality Reduction: Match PCA, naive Bayes, logistic regression with model types (official answer: C) PDF
- Q.19 · MCQ · 1 mark · Clustering (k-Means): Point necessarily in the same k-means cluster as two given points (official answer: D) PDF
- Q.20 · MCQ · 1 mark · Naive Bayes: Number of parameters of a naive Bayes classifier with binary attributes (official answer: B) PDF
- Q.22 · MCQ · 1 mark · Linear Discriminant Analysis: Eigen equation satisfied by the Fisher linear discriminant direction (official answer: A) PDF
- Q.42 · MCQ · 2 marks · Hierarchical Clustering: Dendrogram from single linkage clustering on a distance table (official answer: A) PDF
- Q.43 · MCQ · 2 marks · Neural Networks: Weights making two ReLU networks equivalent for positive inputs (official answer: A) PDF
- Q.53 · MSQ · 2 marks · Support Vector Machines: Which of four plotted two-class datasets are linearly separable (official answer: A, D) PDF
- Q.62 · NAT · 2 marks · Decision Trees: Information gain of an attribute on a cricket match table (official answer: 0.12–0.13) PDF
- Q.63 · NAT · 2 marks · k-Nearest Neighbours: Minimum odd k for which kNN labels a point as square (official answer: 5) PDF
Other Data Science and Artificial Intelligence topics
- Probability and Statistics PYQs
- Linear Algebra PYQs
- Calculus and Optimization PYQs
- Programming, Data Structures and Algorithms PYQs
- Database Management and Warehousing PYQs
- Artificial Intelligence PYQs
FAQ
How many Machine Learning questions come in GATE DA?
In the official papers we tagged: 8 in 2026, 9 in 2025 and 10 in 2024. That is 27 questions in all, about 16% of Data Science and Artificial Intelligence. Weightage: 14 marks a year on average (13–15), about 14% of a 100-mark paper.
Which Machine Learning topics repeat in GATE DA?
Asked in the most papers: Neural Networks, PCA and Dimensionality Reduction, Hierarchical Clustering. The concept table lists every one with its count.
Are these official GATE DA questions?
Yes. Every question is from an official GATE DA paper hosted on our year pages. The one-line summaries are ours; open the PDF linked beside each question for the full text.