MAT9004 · Mathematical Foundations For Data Science And Ai
The Complete Exam Bible · S1 2026Mathematical Foundations for Data Science and AI
Mathematical Foundations for Data Science and AI is the mathematical toolkit a data-science or AI degree assumes — six disjoint maths worlds in one paper: single-variable calculus, linear algebra, multivariable optimisation, combinatorics, probability & Bayes, and graph theory. They don't build on one another the way a normal maths unit does, so you can't coast on a single strength. The final exam is 60% of your grade and a hurdle — you must score at least 45% on the exam itself to pass the unit, whatever your assignment marks. It is closed-book with no calculator, but a formula sheet is provided, so the exam tests one thing: can you execute each method by hand on fresh numbers. This guide teaches every examined technique to that standard — the definition stated plainly, the method on a worked example, and the trap that loses marks.
What MAT9004 covers
Six examined topics → one exam-ready map. Each links to its free chapter guide.
How MAT9004 is assessed
| Component | Weight | Format |
|---|---|---|
| Final exam · hurdle | 60% | Closed book · no calculator · formula sheet provided · must score ≥45% on it to pass the unit (hurdle) |
| Applied-class quizzes | 20% | Five short quizzes across the semester — confirm the exact schedule in your unit guide |
| Assignments | 20% | Two assignments submitted across the semester — confirm the exact split in your unit guide |
Stationary points & the second-derivative test — the calculus staple, step by step
- +1Differentiate: f′(x) = 3x2 − 12x + 9.
- +1Set f′(x) = 0: 3x2 − 12x + 9 = 0 ⇒ x2 − 4x + 3 = 0 ⇒ (x − 1)(x − 3) = 0, so x = 1 and x = 3.
- +1Second derivative: f″(x) = 6x − 12.
- +1Classify x = 1: f″(1) = −6 < 0 ⇒ local maximum. The value is f(1) = 6.
- +1Classify x = 3: f″(3) = +6 > 0 ⇒ local minimum. The value is f(3) = 2.
Key terms
- Stationary point
- A point where the first derivative is zero, f′(x) = 0 — a candidate for a local maximum, local minimum, or a point of inflection. The second-derivative test then settles which it is.
- Gaussian elimination
- The row-reduction method for solving a linear system Ax = b: use elementary row operations to reach an upper-triangular form, then back-substitute. The flagship by-hand technique of the linear-algebra block.
- Gradient
- The vector of first partial derivatives, ∇f = (∂f/∂x, ∂f/∂y). It points in the direction of steepest ascent, and setting ∇f = 0 locates the stationary points of a two-variable function.
- Bayes' theorem
- The rule for reversing a conditional probability: P(A|B) = P(B|A)P(A) / P(B). It is the signature long-answer of the probability block, usually set up with the law of total probability in the denominator.
- Handshaking lemma
- In any graph the sum of the vertex degrees equals twice the number of edges, Σ deg(v) = 2|E|. A direct corollary: every graph has an even number of odd-degree vertices.
MAT9004 FAQ
Is MAT9004 hard?
It is broad rather than deep: six disjoint maths worlds — calculus, linear algebra, multivariable optimisation, combinatorics, probability and graphs — in one exam. The difficulty is coverage and speed: each topic is procedural, but you must execute the method by hand, with no calculator, across all six. You can't coast on one strength, so the trap is leaving a whole world under-revised.
How is MAT9004 assessed?
The final exam is 60% of the unit mark and a hurdle — you must score at least 45% on the exam itself to pass, whatever your assignment marks. The remaining 40% is continuous assessment: applied-class quizzes (about 20%) and assignments (about 20%) across the semester. Confirm this year's exact split and schedule in your unit guide.
What is on the MAT9004 final exam?
All six examined topics: single-variable calculus (derivatives, stationary points, integration), linear algebra (vectors, Gaussian elimination, determinants, eigenvalues), multivariable optimisation (gradient and Hessian), combinatorics (counting rules, binomial, inclusion-exclusion), probability and Bayes (conditional probability, the law of total probability, random variables), and graph theory (degrees, trees, adjacency matrices). The exam is closed-book with no calculator, but a formula sheet is supplied inside the paper.
Do I need a strong maths background for MAT9004?
It is a foundation unit, so it starts from first principles, but it moves fast and assumes comfort with school-level algebra and functions. The work is procedural rather than proof-heavy: the marks reward applying the right technique cleanly by hand, line by line, not abstract derivation.
Is using AskSia for MAT9004 cheating?
No. AskSia is a study reference written in our own words — we host none of your lecturer's files, and Sia teaches you the method to earn the marks; it does not complete or sit your assessments.
How to study for the exam
Because the exam is closed-book, no calculator, and a 60% hurdle that samples all six worlds, the winning move is to drill the recurring procedural chains until they are automatic: differentiate → set f′ = 0 → classify with f″; row-reduce → back-substitute; ∇f = 0 → Hessian test; condition → Bayes; degree sequence → count edges. Don't cram what the formula sheet already gives you — standard derivative tables, counting formulas and distribution facts are supplied — spend the time instead on executing the methods cold on fresh numbers. Show every line: method marks are real, and they are the safest marks when you can't reach for a calculator. Above all, spread your revision so no single world is left blank, because the hurdle is on the whole paper.
