How GATE DA Works
Everything you need to know about the paper before you start studying: structure, marking, weightage, and a study plan that starts where the marks are.
What you'll learn
- What the paper looks like: 65 questions, 100 marks, 3 hours, in your browser
- The three question types and the marking rule that changes how you should answer
- Where the marks actually are — and how much time each subject deserves
- A 12-week plan that puts your freshest study time on the heaviest subjects
Before you learn a single formula, spend ten minutes learning the exam. GATE Data Science & Artificial Intelligence is a young paper — first sat in 2024 — and it asks you to hold an unusually wide net: probability and linear algebra, calculus and optimization, Python and data structures, databases, the whole of machine learning, and a little classical AI. That breadth is intimidating only until you see how the marks are spread; then it stops being a wall and becomes a plan.
So this page is the map before the territory. What the paper looks like, the one marking rule that should change how you answer, where the marks actually live, and a twelve-week order that spends your freshest hours where they pay best. There is no quiz at the end — just the lay of the land, so that every lesson after this one has somewhere to sit.
The paper at a glance
In numbers:
| Questions | Marks | |
|---|---|---|
| General Aptitude (GA) | 10 (5 × 1-mark + 5 × 2-mark) | 15 |
| DA core subject | 55 (25 × 1-mark + 30 × 2-mark) | 85 |
| Total | 65 | 100 |
General Aptitude is common to every GATE paper — English, simple maths, charts, reasoning. The other 85 marks come from the Data Science & AI syllabus.
Marking — the one rule that matters
| Question type | What it is | Wrong answer costs you |
|---|---|---|
| MCQ — multiple choice | exactly one correct option | −1/3 (1-mark) or −2/3 (2-mark) |
| MSQ — multiple select | one or more correct; all-or-nothing | nothing |
| NAT — numerical answer | you type a number | nothing |
The mix of MCQ / MSQ / NAT shifts a little year to year (2026 was roughly 33 / 14 / 18). Only the totals are fixed.
Where the marks actually are
GATE never publishes official per-topic weightage, but counting the questions across the 2024, 2025 and 2026 papers gives a clear picture — and it should drive where your hours go.
Reading the chart:
- Heaviest and reliable. Probability & Statistics is the single largest area most years (and very NAT-heavy). Machine Learning is the biggest named subject and trending up.
- High but moves. Linear Algebra was 10 marks in 2024 and 17 in 2025. Programming & DSA peaked around 21 in 2024 and eased after. Both always matter.
- Small but steady. Calculus and Databases sit near 8 every year — low effort, reliable return. Don’t skip them.
- The wildcard. Artificial Intelligence jumped from about 11 marks (2024) to 3 (2025). Cover it; don’t pour weeks into it.
The three mathematics subjects together — Probability, Linear Algebra, Calculus — are roughly a third to a half of the 85 core marks. Strong maths is the highest-leverage investment you can make.
A 12-week study plan
Part-time. Compress or stretch to your timeline; keep the order — it front-loads the heaviest subjects while making sure each one’s prerequisites come first. In particular the maths (Linear Algebra, Calculus) comes before the Machine Learning that leans on it, so you never meet eigenvectors in PCA or the chain rule in backprop before you’ve actually learned them.
| Weeks | Focus | Why now |
|---|---|---|
| 1 | This page + Probability foundations | Set the frame; start the heaviest subject |
| 2-3 | Probability & Statistics (full) | Highest weight, NAT-heavy — needs the most time |
| 4-5 | Linear Algebra | High weight, and the foundation ML leans on — eigenvectors (PCA), ‖w‖ (SVM), the normal equation (regression) |
| 6 | Calculus & Optimization | The derivatives and chain rule that gradient descent and backprop need |
| 7-8 | Machine Learning | Highest named subject, rising — and now you already have the eigenvectors and derivatives it relies on |
| 9 | Programming, Data Structures & Algorithms | High weight; goes quickly once you know Python |
| 10 | Databases & Warehousing + General Aptitude | Compact, high marks-per-hour |
| 11 | Artificial Intelligence | Variable weight; cover, don’t over-invest |
| 12 | Full mocks + revision sheets + weak-spot repair | Simulate exam day, fix what’s shaky |
How this track is built
Each subject lesson follows the same shape: a small “How GATE tests this” panel (weight, frequency, question types), the intuition, a diagram or interactive thing to play with, the exact way GATE phrases the question, a worked previous-year problem, the traps that catch people, and a short GATE-level quiz — including the typed numeric answers (NAT) you’ll face on exam day.
There’s one page that isn’t a lesson: the Review & Mastery Checks page. Once you’ve finished a subject, come back a week later and take its interleaved mastery check — a mix of questions from across the whole subject, out of order. That spaced, mixed retrieval is what makes a topic stick for exam day, long after you first learned it. Don’t skip it; it’s where retention actually happens.
You do not have to march through in strict order — but if you are unsure where to stand, stand at the start of the heaviest subject. Begin with Probability & Statistics; the rest of the track points back whenever you reach a prerequisite you have not met yet.
Which is exactly where we begin. Before you can compute a single probability you need a way to count — the outcomes in an event, and the outcomes in the whole sample space — and counting, done with care, turns out to be its own small art. The next page is Counting: the quiet toolkit every probability in this course is built on. Good luck.