UniMelb · MAST90105 · Methods of Mathematical Statistics

MAST90105: pass the exams, not just read the notes

Your complete guide to University of Melbourne's methods of mathematical statistics unit. See where the marks are, work real practice questions, and study with an AI tutor that knows MAST90105.

12.5 credit points Postgraduate Offered S1 ~80% exams School of Mathematics and Statistics

Sia generates MAST90105 practice questions, works through them step by step, and quizzes you on the material the exam weights most heavily.

Try a real exam-style question

Worked example

Multiple choice · solution revealed after you answer

You observe n independent Bernoulli(p) trials with k successes. What is the maximum likelihood estimator of p?

Worked solution

The likelihood is L(p) = p^k (1 − p)^(n − k).

Take the log-likelihood: l(p) = k ln p + (n − k) ln(1 − p).
Differentiate and set to zero: k/p − (n − k)/(1 − p) = 0.
Solving gives p_hat = k / n, the sample proportion of successes.

The trap: Guessing p_hat = k (the raw count) or inverting to n/k. The maximum likelihood estimator maximises the log-likelihood, which gives the sample proportion k/n; the (k+1)/(n+2) form is a Bayesian posterior mean, not the MLE. classic slip!

your whole grade
Where your grade comes from Exams 80% · Assignment 20%

One exam decides 35% of your grade. This whole page is built around that.

Overview

What MAST90105 is, and where it sits

MAST90105 Methods of Mathematical Statistics is the University of Melbourne's postgraduate mathematical-statistics subject, taught in the School of Mathematics and Statistics. It builds statistical inference from its probability foundations: probability and Bayes' theorem, discrete and continuous random variables and moment-generating functions, the central limit theorem, bivariate distributions and correlation, transformations and sampling distributions, point estimation by the method of moments and maximum likelihood, estimator properties and the Cramér-Rao lower bound, interval estimation via pivotal quantities, hypothesis testing and the Neyman-Pearson lemma, and distribution-free and categorical methods.

The subject is theoretically rigorous and quantitative, with an applied component in R. Most of the grade is exam-based, rewarding the ability to derive estimators and tests and prove their properties rather than apply canned procedures. The recurring skill is moving fluently between probability theory, estimation and the formal logic of hypothesis testing.

How it differs from its first-year siblings. Methods of Mathematical Statistics is the theory subject: it derives why estimators and tests work (maximum likelihood, the Cramér-Rao bound, Neyman-Pearson) rather than just applying statistical recipes.

Official outline: handbook.unimelb.edu.au · MAST90105 outline. Always treat the official outline and the exam timetable as authoritative.

Difficulty & time commitment

Is MAST90105 hard, and how much time does it take?

MAST90105 is manageable if you keep a weekly rhythm and treat the back half as the main event. The pattern is consistent: it starts gently and steepens, and the heaviest assessment is the part that separates grades.

Difficulty
3.5 / 5
Moderate–Hard. Gentle early, demanding back half. Hard to fail with steady work; a top grade takes consistent practice.
Exam load
80%
The exams decide most of the grade. The heaviest single component is 35%.
Weekly time
~10 hrs
Around 10 hours per week including class, across lectures, study and assessment.
Probability and distributionsbuilds the theory
Estimation and inferencesteep (MLE, Cramér–Rao, Neyman–Pearson)

The difficulty curve and the assessment weighting point the same way: the back half is harder and worth more. Front-loading effort there is the highest-return decision in the unit.

Is this unit for you

Who tends to do well, and who tends to struggle

You will likely do well if

  • You are comfortable with probability theory, calculus and formal proof, since the subject derives rather than applies.
  • You practise deriving estimators and tests by hand, not just running them in R.
  • You keep pace through the estimation-and-inference block, the steepest part.

You may struggle if

  • You are shaky on probability and calculus, which the estimation theory assumes.
  • You treat statistics as recipes; this subject asks you to prove why methods work.
  • You under-practise the maximum-likelihood, Cramér-Rao and Neyman-Pearson material.
do this ↘
What top students do differently
  • Derive the standard MLEs (Bernoulli, normal, Poisson) from the log-likelihood until it is routine.
  • Practise the Cramér-Rao bound and Neyman-Pearson lemma as proofs, not formulas.
  • Work past exams by hand and use R only to check, since the exams reward derivation.

Syllabus

The 10 topics, topic by topic

The exam-weight marker on each topic shows where the marks concentrate. The amber topics carry the highest exam weight.

T1 · Probability Foundations and Bayes

Axioms & set identities · counting · conditional probability · total probability & <b>Bayes’ theorem</b> · independence

Lower exam weight

T2 · Discrete Random Variables and MGFs

PMF / CDF · expectation & variance · <b>moment generating functions</b> · the discrete families

Lower exam weight

T3 · Continuous Random Variables and the CLT

Density / CDF · mean, median, variance · exponential / gamma / χ² · <b>normal</b> · CLT & LLN

Lower exam weight

T4 · Bivariate Distributions and Correlation

Joint / marginal / conditional · covariance & <b>correlation</b> · the <b>bivariate normal</b> · zero-corr ≠ independence

Lower exam weight

T5 · Transformations and Sampling Distributions

<b>CDF method</b> · Jacobian / change of variables · <b>MGF method</b> · <b>Z → t → χ² → F</b>

Lower exam weight

T6 · Point Estimation: Method of Moments and Maximum Likelihood

<b>Method of moments</b> · the likelihood · <b>maximum likelihood</b> — the full L → ln L → score = 0 recipe

Lower exam weight

T7 · Estimator Properties and the Cramér–Rao Lower Bound

Bias, variance & <b>MSE</b> · <b>Fisher information</b> · the <b>Cramér–Rao lower bound</b> · regularity conditions · asymptotic normality

Lower exam weight

T8 · Interval Estimation and Pivotal Quantities

<b>Pivotal quantities</b> → CIs for means & proportions · large-sample CIs · distribution-free CIs for percentiles

Lower exam weight

T9 · Hypothesis Testing and Neyman–Pearson

Type I / II & power · the <b>Neyman–Pearson</b> lemma & LRTs · one-sample <b>t</b>, variance <b>χ²</b>, proportion <b>z</b>

Lower exam weight

T10 · Distribution-Free and Categorical Methods

Sign & rank tests · <b>goodness-of-fit χ²</b> · <b>contingency tables</b> & tests of independence

Lower exam weight

How it's assessed

Assessment structure

ComponentWeightFormat & timing
Mid-semester exam35%Mid-semester · on-campus, invigilated, <b>3 hours</b>; the <b>probability half</b> (Weeks 1–7) · <b>one A4 double-sided sheet</b> + Casio FX-82 · <b>distribution table provided</b>.
Final exam35%End of semester · on-campus, invigilated, <b>3 hours</b>; cumulative, weighted to the <b>inference half</b> (Weeks 8–12) · <b>one A4 sheet</b> + Casio FX-82 · <b>distribution table provided</b>.
Written assignments20%Four individual written assignments across the semester (~5% each).
Computer / R lab test10%In a lab on a laptop, <b>open-book</b>; R-based; no communication — the only open-book task.
Mid-semester exam35%
Mid-semester · on-campus, invigilated, <b>3 hours</b>; the <b>probability half</b> (Weeks 1–7) · <b>one A4 double-sided sheet</b> + Casio FX-82 · <b>distribution table provided</b>.
Final exam35%
End of semester · on-campus, invigilated, <b>3 hours</b>; cumulative, weighted to the <b>inference half</b> (Weeks 8–12) · <b>one A4 sheet</b> + Casio FX-82 · <b>distribution table provided</b>.
Written assignments20%
Four individual written assignments across the semester (~5% each).
Computer / R lab test10%
In a lab on a laptop, <b>open-book</b>; R-based; no communication — the only open-book task.
  • Pass on a weighted average of at least 50%. No single-component hurdle unless noted; confirm against the official subject page.
read this! If you read nothing else

This is an exam-cram unit. With the exams at 80% of the grade and the mid-semester exam alone at 35%, your result is overwhelmingly decided by how well you perform under time pressure.

How to actually pass it

A weekly rhythm, two checklists, and the traps to avoid

The unit rewards consistency over cramming, and practice over re-reading. Here is the loop that works, then what to have nailed before each exam.

The weekly loop

Before lecture
Review the prior probability or estimation result so each new derivation extends a solid base.
Each assignment
Derive results by hand and verify numerically in R, understanding both.
Weekly
Maintain a derivations sheet (key MLEs, distributions, test statistics) you can reproduce.

Before the mid-semester checklist

Before the final heaviest topics

  • Rehearse maximum likelihood and method-of-moments estimation for standard distributions.
  • Drill the Cramér-Rao lower bound and estimator-property proofs.
  • Practise Neyman-Pearson hypothesis tests and interval estimation via pivotal quantities.
  • Work both exams by hand under time pressure across probability and inference.

The mistakes that cost marks

01

Applying recipes without deriving. The subject rewards derivation and proof. Running R procedures without understanding the theory leaves you unable to answer the derivation questions that carry the marks.

02

Confusing MLE and Bayesian estimates. The MLE maximises the likelihood (k/n for Bernoulli); posterior means like (k+1)/(n+2) are different estimators. Mixing them is a common inference error.

03

Cramming the inference block. MLE, Cramér-Rao and Neyman-Pearson are the steepest, most heavily examined topics and do not compress into the final week.

Teaching team

Who teaches MAST90105

No teaching staff are publicly listed for this offering. Check the official course page for the current coordinator and lecturers.

Formula & concept sheet

The vocabulary and formulas you must own

Likelihood and MLE
The likelihood L(theta) is the joint density viewed as a function of the parameter; the maximum likelihood estimator maximises L (or its log). For Bernoulli(p) it gives k/n.
Method of moments
Estimates parameters by equating sample moments to population moments and solving; a simpler alternative to maximum likelihood.
Cramér-Rao lower bound
A lower bound on the variance of any unbiased estimator, equal to the inverse of the Fisher information; an estimator achieving it is efficient.
Neyman-Pearson lemma
For testing two simple hypotheses, the likelihood-ratio test is the most powerful test at a given significance level.
Central limit theorem
The sample mean of n independent identically distributed variables is approximately normal for large n, regardless of the underlying distribution.

Common acronyms: MLE · MoM · CLT · MGF · CRLB · NP · iid.

Where it fits

Prerequisites, related units & why it matters

Postgraduate mathematical-statistics subject; assumes probability, calculus and prior statistics. Check the UniMelb Handbook for prerequisites.

Why it matters beyond the grade. The rigorous inference foundation underpins statistics, data science, actuarial and quantitative-research roles where deriving and justifying methods matters.

FAQ

Frequently asked questions

Is MAST90105 hard?

It is moderate-to-hard: a rigorous postgraduate mathematical-statistics subject that derives estimators and tests and proves their properties. Strong probability and calculus and consistent hand derivation make it manageable, but it is theory-heavy and exam-weighted.

How is MAST90105 assessed?

A 35% mid-semester exam, a 35% final exam, written assignments worth 20%, and a computer/R lab test worth 10%. The components sum to 100% and the two exams dominate.

What does it cover?

Probability and Bayes, random variables and moment-generating functions, the central limit theorem, estimation by method of moments and maximum likelihood, the Cramér-Rao bound, interval estimation, Neyman-Pearson hypothesis testing, and distribution-free methods.

How much maths is involved?

A lot: it is mathematical statistics, requiring probability theory, calculus and formal derivation and proof, with an applied component in R.

What background do I need?

Solid probability, calculus and ideally prior statistics. The subject builds inference rigorously from probability foundations.

Study MAST90105 with Sia

Work through the core topics and the rest of the unit with a tutor that knows it and quizzes you on the topics the assessments weight most heavily.

Start studying with Sia