Last updated: 2026-10-09

CMI Data Science Syllabus and PYQ Weightage 2026

The CMI MSc Data Science entrance exam syllabus is heavily anchored in School Level Mathematics, which commands 37.2% of all previous year questions, followed closely by Probability Theory at 22.87%. To secure an interview call, you must master these two core areas while strategically navigating speed traps like Combinatorics. This forensic weightage guide breaks down the exact 2018 to 2026 PYQ data, the 3.5 hour exam pattern, and the raw score benchmarks you need to hit the 99 percentile equivalent.

What is the CMI Data Science exam pattern?

The CMI MSc Data Science entrance exam runs for 3.5 hours (2:00 PM to 5:30 PM) in a single offline sitting. It contains 40 questions worth 100 marks, split into Part A (objective) and Part B (descriptive), testing math, probability, discrete math, and programming.

The 3.5 hour duration is a deliberate choice by the institute. While the MSc and PhD Mathematics, Computer Science, and Physics papers run for exactly three hours, the Data Science paper gets an extra thirty minutes. This extension exists purely to accommodate the larger descriptive component in Part B. You are not just selecting multiple choice options. You are writing out mathematical proofs, deriving probability distributions, and dry running algorithms with a pen on paper.

Part A typically consists of 20 objective questions. These are designed to test your breadth and speed. Objective sections typically carry a penalty for incorrect guesses, making blind guessing a negative expected value strategy. Part B contains the remaining 20 questions, which are descriptive. This is where the real filtering happens. The evaluators at CMI look for clarity of thought, rigorous logical steps, and mathematical elegance. A correct final answer with no working will fetch you zero marks in Part B. You must show every step of your derivation.

The exam is conducted offline in pen and paper mode across roughly 37 exam centres in India. You need to build the physical stamina to write continuously for 210 minutes. Many candidates from pure computer science backgrounds struggle with the sheer volume of handwriting required to prove discrete math theorems or statistical properties. Practice your mock tests with a physical notebook and a timer to condition your hand and your brain for the actual exam environment. A recommended time split is 60 minutes for Part A, 140 minutes for Part B, and 10 minutes for final review.

Which subject has the highest weightage in CMI Data Science?

School Level Mathematics heavily dominates the CMI Data Science syllabus with 37.2% of all previous year questions (109 out of 293 total). Probability Theory follows closely at 22.87%, making these two subjects the absolute core for securing an interview call.

Let us break down the 293 previous year questions we analyzed from 2018 through 2026. The distribution is not random. It reflects the exact mathematical foundations the institute believes are necessary for advanced data science research. School Level Mathematics is the undisputed king of this paper. This includes polynomials, matrices, determinants, linear equations, and basic function properties. Do not mistake school level for easy. CMI takes standard algebraic concepts and twists them into complex, multi step logical puzzles. Matrices and Linear Systems alone account for 4.44% of the total paper, while Determinants and Matrix Inverses add another 3.75%. If your matrix algebra is weak, you will fail the linear algebra questions later in the paper.

Probability Theory is the second pillar, holding 22.87% of the weightage. Data science is fundamentally applied probability. The exam heavily tests Conditional Probability, Bayes Theorem, and Independence, which together make up 4.1% of the paper. You must be comfortable deriving expected values and variances for discrete random variables (2.73%) from first principles. Rote memorization of formulas will not work here. You need to understand the underlying sample spaces and counting mechanisms.

Discrete Mathematics captures 21.16% of the paper. This is the bridge between pure math and computer science. Topics like sets, relations, combinatorics, the pigeonhole principle, and graph counting are heavily tested. Graph Counting, Handshakes, and Tournaments alone make up 2.39% of the exam. You will be asked to prove properties of graphs or count the number of valid paths under strict constraints. Logical Reasoning and Truth Teller Puzzles also feature prominently at 2.39%, testing your ability to parse complex boolean conditions.

Programming and Algorithmic Thinking make up 11.95% of the paper. Notice that this is not a coding test in the traditional sense. You will not be writing Python or C++ syntax on an IDE. You will be given pseudocode or logical descriptions of Array Processing and Linear Scan Algorithms (2.39%), and you must dry run them to find the output or the time complexity. Searching and Selection Algorithms (2.05%) also feature prominently, requiring you to trace loop iterations and state changes manually.

CMI Data Science Marks vs Percentile Benchmarks

Securing an interview call at the Chennai Mathematical Institute requires a raw score between 45 and 55 out of 100, while hitting the 99 percentile equivalent demands 75 or more marks. These benchmarks are derived from historical paper difficulty and slot normalization variations.

The raw score required to clear the CMI Data Science cutoff is surprisingly low compared to other national level entrances. A score between 45 and 55 out of 100 is generally sufficient to secure an interview call. Why is the cutoff so low? The answer lies in the brutal grading of Part B. The descriptive questions are designed to be extremely challenging. The institute is not looking for candidates who can solve 40 routine problems. They are looking for candidates who can solve 15 highly complex problems with absolute mathematical rigor.

To hit the 45 to 55 band, you need to maximize your accuracy in Part A. Aim for 12 or more correct objective questions. Then, in Part B, you only need to fully solve and prove 3 or 4 descriptive questions. Do not attempt to write half baked answers for all 20 Part B questions. The evaluators will penalize you for logical leaps. Pick the questions where you can write a complete, unbroken chain of logic from premise to conclusion. To hit the 75 plus band, you need near perfect accuracy in Part A (17 or more correct) and the ability to crack at least 5 complex descriptive proofs in Part B.

What is the eligibility for CMI MSc Data Science?

CMI strictly requires an undergraduate degree (B.A., B.Sc., B.Math., B.Stat., B.E., B.Tech. or equivalent) with a background in Mathematics, Statistics, or Computer Science. Final-year undergraduates expecting to complete their degree by the academic start date are also eligible to apply.

There is a persistent myth that the Chennai Mathematical Institute only accepts students with pure mathematics or statistics degrees. This is entirely false. The official eligibility criteria explicitly welcome B.E. and B.Tech. graduates, provided they have a demonstrable background in Mathematics, Statistics, or Computer Science. The institute deliberately keeps the wording broad to attract quantitative engineers, physicists, and other analytical graduates. If you are an engineering student, your coursework in linear algebra, probability, and data structures perfectly aligns with the CMI syllabus.

The key transition for engineers is moving from applied mathematics to proof based mathematics. Engineers are used to calculating numerical answers to build systems. CMI requires you to show why a theorem holds true for all possible cases. You must learn to write proofs, use induction properly, and structure your logical arguments. Final year undergraduates who expect to complete their degree by the start of the academic session are fully eligible to apply and sit for the exam.

Exam Schedule and Live Dates

The admission cycle follows a strict annual rhythm. The is the primary target for your preparation. The has already concluded, and its papers are now part of the official archive.

Most candidates treat the CMI Data Science exam like a standard university semester test. They grind LeetCode for programming and ignore the foundational school-level algebra that actually makes up over a third of the paper. CMI does not care if you can invert a binary tree in optimal time if you cannot elegantly solve a system of linear equations or apply the pigeonhole principle. The institute selects for mathematical maturity, not just software engineering syntax.

Your Next Steps and Study Resources

Do not waste time gathering scattered materials. Your preparation must be focused entirely on the high-yield areas identified in our forensic weightage analysis. Start by downloading the official syllabus breakdown, then move immediately into solving previous year papers under strict timed conditions. Use the chapter practice modules to drill your weak areas identified by the AI.

Frequently Asked Questions