IIT Madras BS, Diploma in Data Science
What a senior would tell you before the machine learning courses: what each one really asks of you, how they fit together, how every mark is counted, and how to get through the quizzes, OPPEs, Kaggle assignments and the project viva. Checked against the official course pages and the grading document.
Not an official course page. Rules and dates change every term, so check your term's grading document. No answers to graded work here, only how to approach it.
Three courses and a project
14 credits
MLF 4, MLT 4, MLP 4 and the MLP Project 2.
Official lectures
About 92 h
MLF 37 h, MLT 27 h, MLP 28 h, all free on YouTube.
Sep 2026 term
Starts 2 Oct
Quiz 1 on 15 Nov. End term on 10 Jan 2027.
MLP is different
No quizzes
Two OPPEs and three Kaggle assignments instead.
Three courses, three different skills
Calculus, linear algebra, optimisation and probability. Nearly every question is a calculation by hand, so practise problems every single week.
PCA to SVMs to neural networks. Know what each model assumes, what it minimises and how it trains. Derive it; don't memorise it.
The same models in scikit-learn, marked on code: two OPPEs and three Kaggle assignments. Your numbers have to match exactly.
| Date | What | For |
|---|
MLF, MLT and MLP share the letters ML and almost nothing else. One is maths, one is theory, one is code. They reward different habits, and most of the pain in these courses comes from studying all three the same way.
The project is a separate two-credit course with its own registration. Its fee stays valid for two terms.
Together with the MLP Project they make up most of the Diploma in Data Science. Here is what each one is, straight from the official course pages and the grading document.
The maths under machine learning: calculus, linear algebra, optimisation and probability. No model is trained; nearly every question is a calculation.
The algorithms, derived: PCA, clustering, regression, Naive Bayes, SVMs, ensembles and neural networks, and why each one works.
The same models in Python with scikit-learn, on real datasets. Marked mostly on code, not on paper.
A Kaggle competition among students. Beat a score cutoff from one registered notebook, then defend your work in two vivas.
I took these courses in this programme: an S grade in MLT with 100/100 in the end term, and 11th of 2,593 on the Kaggle leaderboard of the MLP course's competition. Everything below is either from an official page, linked where it is used, or clearly marked as my own experience.
There is only one hard rule: MLT before MLP. Everything else can share a term.
MLF is a co-requisite of MLT: take it before MLT or in the same term. MLT is a prerequisite of MLP: finish it first. MLP is a co-requisite of the MLP Project and of Tools in Data Science.
Source: requisites table, grading document, May 2026. The MLP course page lists MLF and MLT both as prerequisites; the grading document is newer.
The grading document gives its own plans for someone doing only the Data Science diploma. Here are the three that use the Business Analytics track; the Deep Learning track swaps BA for the Deep Learning and Generative AI course and its project. All three agree: MLF first, MLP after MLT, and TDS in the same term as MLP.
| Plan | Term 1 | Term 2 | Term 3 | Term 4 |
|---|---|---|---|---|
| 2 terms "not recommended unless you have some background" | MLF, MLT, BDM, BDM project | MLP, BA, TDS, MLP project | – | – |
| 3 terms "moderate pathway" | MLF, BDM, BDM project | MLT, BA | MLP, TDS, MLP project | – |
| 4 terms "extremely comfortable and relaxed" | MLF, BDM | MLT, BDM project | MLP, TDS | BA, MLP project |
Week 4 of MLF is eigenvalues. Watch it in your first week of MLT.
MLF and MLT together, the most common pairing, starts out of sync. MLT opens with PCA in Week 1. The MLF lectures only reach eigenvalues in Week 4 and PCA in Week 7. For about a month MLT uses maths that MLF has not taught you yet. Watch the MLF eigenvalue lectures early, or 3Blue1Brown's short chapter on eigenvectors, and the gap closes.
MLP and its project in the same term is allowed, but the guideline warns against it: "If you are new to ML, you should complete the MLP course in one term and the MLP project in a later term." The project's milestones don't follow the course's week order, so you would be learning some topics before the lectures reach them.
None of the three courses teaches its prerequisites again. MLF opens with one week of calculus and then goes straight to subspaces. MLP opens with pandas and assumes you can already write Python.
| Be comfortable with | Learned in | First needed in |
|---|---|---|
| Matrices, matrix products, solving Ax = b | Maths 1 and 2 | MLF Week 3, MLT Week 1 |
| Derivatives, partial derivatives, the gradient | Maths 1 and 2 | MLF Week 2, every loss in MLT |
| Eigenvalues of a small matrix | Maths 2 | MLT Week 1, MLF Week 4 |
| Probability, expectation, the normal distribution | Stats 1 and 2 | MLT Week 4, MLF Weeks 10–12 |
| Python functions, loops, lists, dictionaries | Python | MLP Week 1, MLT programming assignments |
| Reading and filtering a table in pandas | Your own practice | MLP Week 1 (the whole week is pandas) |
Tick what is true today. It stays on this device only.
Fixing the gaps. For the maths lines, 3Blue1Brown's Essence of Linear Algebra and Essence of Calculus are the fastest way back in. For probability, Seeing Theory. For pandas, the free course on Kaggle Learn takes an evening.
A maths course wearing an ML name tag. In the course page's own words, it covers "fundamentals that do not necessarily fall under Machine Learning but are quite necessary" for it.
93 lectures, about 37 hours. Every one is on YouTube.
Calculus, linear algebra, optimisation and probability, each tied back to a machine learning problem. Taught by Harish Guruprasad Ramaswamy, Arun Rajkumar and Prashanth L.A. Here is the course week by week, from the titles in the official lecture playlist. The course page orders the middle weeks a little differently; follow your portal.
| Week | What you learn | Where it comes back |
|---|---|---|
| 1 | What ML is: regression, classification, dimensionality reduction, density estimation | The map for the whole diploma |
| 2 | Calculus: continuity, derivatives, linear approximation, many variables | Every gradient in MLT |
| 3 | The four fundamental subspaces, projections, least squares | Linear regression is a projection |
| 4 | Linear and polynomial regression; eigenvalues, eigenvectors, diagonalisation | PCA |
| 5 | Complex, Hermitian and unitary matrices | Why symmetric matrices behave |
| 6 | Singular value decomposition; positive definite matrices | PCA, compression |
| 7 | Principal component analysis, as maximising variance | MLT Weeks 1–2 |
| 8 | Unconstrained optimisation, gradient descent, Taylor series | How every model trains |
| 9 | Constrained optimisation, Lagrange multipliers, convexity | SVMs, regularisation |
| 10 | Convexity in ML; primal and dual; KKT conditions | The SVM dual in MLT |
| 11 | Continuous random variables; uniform, exponential, normal | MLE, Naive Bayes |
| 12 | Multivariate normal, estimation, Gaussian mixtures and EM | MLT Week 4 |
Weeks 3 to 6, the linear algebra block. It moves fast and builds on itself: a doubt left over from Week 3 becomes three doubts by Week 6. Week 5's Hermitian and unitary matrices feel far from ML until you see they are the reason PCA and the SVD work.
The optimisation weeks are where it turns into ML. Gradient descent, convexity, Lagrange multipliers and KKT are exactly the tools MLT uses for regression, logistic regression and SVMs. Time spent here pays twice.
Almost every question is a number. One slip in the last line costs the whole question, so checking your arithmetic is a skill worth practising on its own.
Free book that matches the course: Mathematics for Machine Learning, chapters 2 to 7.
Books that match. Chapters 2–7 of Mathematics for Machine Learning follow MLF almost chapter for chapter, and it is free. For convexity and KKT, Boyd and Vandenberghe is the standard text, also free. The subspace weeks are Gilbert Strang's own framing, so his MIT course is the natural companion.
T = 0.05·GAA + max( 0.6·F + 0.25·max(Qz1, Qz2), 0.4·F + 0.25·Qz1 + 0.3·Qz2 )
GAA is the average of your best 10 of 11 scores (Week 1–10 assignments and Mock Quiz 1). Qz1, Qz2 are the quiz scores, 0 if you missed one. F is the end term. All out of 100. From the May 2026 grading document.
The grader takes the better of two formulas. The first counts only your better quiz, so one bad or missed quiz does not sink you. The second rewards doing well in both and gives Quiz 2 more weight. You never choose; the max is taken for you.
Worked example. GAA 80, Quiz 1 50, Quiz 2 70, end term 60. First formula: 36 + 17.5 = 53.5. Second: 24 + 12.5 + 21 = 57.5. So T = 4 + 57.5 = 61.5. Try your own numbers in the calculator.
To sit the end term: the average of your best 5 of 8 scores (Week 1–7 assignments and Mock Quiz 1) must be at least 40, and you must attend at least one of the two quizzes. The same rule applies to MLT.
Source: grading document, MLF and MLT sections, updated 15 July 2026. Mock Quiz 1 shows up on your dashboard as "Mock Assignment".
Their assignments are not part of GAA, so it is tempting to skip them. Their content is in the end term, and the grading document says so in as many words. Do them.
The course where the maths becomes algorithms you can explain. It is also the course the MLP Project viva leans on most.
74 lectures, L1 to L74, about 27 hours.
Taught by Arun Rajkumar. The course page sets the aim: "the main methods and models used in machine learning problems of regression, classification and clustering", and "their suitability for different problems". Its week list matches the lecture order.
| Week | What you learn | Lectures (my grouping) |
|---|---|---|
| 1 | Representation learning and PCA | L1–L9 |
| 2 | Kernel PCA | L10–L12 |
| 3 | Clustering: K-means and kernel K-means | L13–L18 |
| 4 | Estimation: MLE, Bayesian estimation, Gaussian mixtures and EM | L19–L26 |
| 5 | Regression: least squares and its Bayesian view | L27–L35 |
| 6 | Ridge and LASSO | L36–L40 |
| 7 | Classification: k-nearest neighbours, decision trees | L41–L44 |
| 8 | Generative models: Naive Bayes | L45–L51 |
| 9 | Perceptron and logistic regression | L52–L56 |
| 10 | Support vector machines | L57–L65 |
| 11 | Ensembles: bagging and boosting (AdaBoost) | L66–L71 |
| 12 | Neural networks; multiclass classification | L72–L74 |
For every method, be able to say five things: what data it expects, what model it can represent, what loss it minimises, how it is trained, and how it is evaluated. This is not my framing. It is exactly what the MLP Project's Level 2 viva asks for, for "every machine learning model that you have learnt and/or used", and the programme even has a short official series on it called MLT to MLP. Here is a start. Write your own as the course goes.
| Method | Model | Loss or objective | Trained by |
|---|---|---|---|
| PCA | Projection onto the top k eigenvectors of the covariance | Reconstruction error (same as maximising variance) | Eigendecomposition |
| Kernel PCA | PCA in a feature space given by a kernel | Same, in feature space | Eigendecomposition of the kernel matrix |
| K-means | k centroids; each point joins the nearest | Within-cluster sum of squared distances | Lloyd's algorithm: assign, update, repeat |
| Gaussian mixture | Weighted sum of k Gaussians | Negative log-likelihood | EM |
| Linear regression | w·x | Squared error | Normal equations or gradient descent |
| Ridge / LASSO | w·x | Squared error + λ‖w‖² or + λ‖w‖₁ | Closed form (ridge); subgradient methods (LASSO) |
| k-NN | Vote of the k nearest training points | None, nothing is fitted | Store the data |
| Decision tree | Axis-aligned splits into regions | Impurity (entropy or Gini) at each split | Greedy splitting |
| Naive Bayes | Features independent given the class | Likelihood | Counting, with Laplace smoothing |
| Perceptron | sign(w·x) | Number of mistakes | Update w on each mistake |
| Logistic regression | σ(w·x), a probability | Log loss | Gradient descent |
| SVM | sign(w·x + b) with the widest margin | Hinge loss + ‖w‖² | The dual problem; kernels for curved boundaries |
| Bagging, random forest | Average of trees on bootstrap samples | Each tree's own | Trees trained independently |
| AdaBoost | Weighted vote of weak learners | Exponential loss | Reweight the misclassified points each round |
| Neural network | Layers of weighted sums and non-linearities | Cross-entropy or squared error | Backpropagation and gradient descent |
Bishop's chapter map: regression 3, classifiers 4, neural nets 5, SVMs 7, EM 9, PCA 12, boosting 14.
The course page prescribes three books: Duda, Hart and Stork's Pattern Classification, Bishop's Pattern Recognition and Machine Learning, and The Elements of Statistical Learning. The last two are free from their authors. If they feel heavy, An Introduction to Statistical Learning is the gentle version, and Stanford's CS229 notes derive most of the same algorithms in a few pages each. Closest of all are notes written for this course: the MLT notes on the programme's GitHub, with slides and Colab notebooks for each week.
T = 0.05·GAA + max( 0.6·F + 0.25·max(Qz1, Qz2), 0.4·F + 0.25·Qz1 + 0.3·Qz2 )
The same formula, GAA and eligibility as MLF. Plus a bonus of 3 marks for submitting the programming assignments, if your average across all assignments is at least 40.
Like every bonus in the programme, it is added only if you pass. It can lift your grade; it cannot turn a fail into a pass.
The MLT course team also runs its own YouTube channel with weekly TA sessions, instructor sessions, workshops and a past-paper session every term.
The same models, now in scikit-learn. It is a programming course in the way it is marked, not only in what it teaches: 60% of the course is code.
98 videos, about 28 hours, many of them notebooks run live on California housing, MNIST and MovieLens.
Taught by Ashish Tendulkar of Google Research. The course page calls it the "companion course to the ML Theory course". The week list on that page no longer matches the lectures (it shows Graph Theory in Week 2), so this one comes from the lecture playlist.
| Week | What you learn | In scikit-learn |
|---|---|---|
| 1 | pandas: Series and DataFrames, selecting, dropping, summaries | before any model |
| 2 | Loading data, missing values, encoders, scaling, outliers, feature selection, PCA | SimpleImputer, OneHotEncoder |
| 3 | Pipelines and composite transformers; class imbalance | Pipeline, ColumnTransformer |
| 4 | A tour of the models, evaluation, regularisation, tuning | GridSearchCV |
| 5 | Regression end to end on California housing | SGDRegressor |
| 6 | Classification on MNIST: perceptron to SVC to ensembles | LogisticRegression, SVC |
| 7 | Large-scale learning; confusion matrix, precision, recall, F1 | classification_report |
| 8 | Not in the public playlist; see your portal | |
| 9 | Clustering; an end-to-end project from framing to fine-tuning | KMeans |
| 10 | Recommender systems: content-based, collaborative filtering, matrix factorisation | MovieLens |
| 11 | Time series in pandas: resampling, decomposition, stationarity | pandas |
| 12 | Transfer learning; neural networks recap |
T = 0.1·GAA + 0.3·F + 0.2·OPPE1 + 0.2·OPPE2 + 0.2·KA
GAA as in MLF (best 10 of Weeks 1–10 and Mock Quiz 1). OPPE1 and OPPE2 are the two online proctored programming exams. KA is the average of the three Kaggle assignments. From the May 2026 grading document.
You can sit the MLP end term if your best 5 of 8 (Weeks 1–7 and Mock Quiz 1) average at least 40; no quiz is needed. To get a grade, you must sit the end term and score at least 40/100 in at least one OPPE. OPPE 1 is only scheduled if you finish the System Compatibility Test first.
Source: grading document, MLP section and OPPE eligibility.
You don't always repeat the whole course. The grading document's own table:
| OPPEs | End term | Total T | What you get |
|---|---|---|---|
| Absent or not eligible | Absent or not eligible | – | U, repeat the course |
| Missed both | Present | 35 or more | I_OP, take the OPPEs alone next term |
| Missed both | Present | below 35 | U |
| Best OPPE below 40 | Present | 40 or more | I_OP, take the OPPEs alone next term |
| Best OPPE below 40 | Present | below 40 | U |
| Best OPPE below 40 | Absent | – | I_BOTH, OPPEs and end term next term |
| Best OPPE 40 or more | Absent | – | I, the end term alone next term |
| Best OPPE 40 or more | Present | – | Graded on T |
Each KA is a small Kaggle competition with a notebook you submit and a peer review after it. In May 2026 they ran 10–22 July, 27 July–4 August and 10–18 August. Together they are 20% of the course, as much as either OPPE.
From my own May 2026 assignments. Approach only; the datasets change.
Three habits mattered more than the choice of model.
submission.csv.A real Kaggle competition, 12 to 15 weeks long, where everyone works alone on the same problem. Beat the cutoff, then pass two vivas with an industry examiner.
Everything here is from the May 2026 guideline. Read your own term's the day it comes out: the cutoff, the dates and the notebook name change.
The problem is a regression or a classification task, new every term. Rules below are from the MLP Project Guideline for May 2026.
Most project trouble in the first weeks is registration trouble. The guideline's steps, in order:
rollno-notebook-term, all lowercase. May 2026's example: 21f1001234-notebook-2026t2.iitmbscs2008p (view access is enough) and keep it private.| Rule | What it means in practice |
|---|---|
| Only the FinalNotebook counts | A CSV you upload by hand, or a submission from any other notebook, does not count. |
| At least three different models | One model only disqualifies the project. |
| No external data | Train only on the competition's files. |
| CPU only | Do not switch on Kaggle's GPU or TPU. |
| Keep it private | Code found public means disqualification. |
| Allowed libraries only | NumPy, pandas, Matplotlib, scikit-learn, XGBoost, Seaborn, imbalanced-learn, SciPy, regex, LightGBM, Plotly, statsmodels, WordCloud, CatBoost and Python's built-ins. Not TensorFlow, PyTorch, NLTK, pickle, AutoML tools or spaCy. |
| No plagiarism | Sharing notebooks or snippets, copying anyone, or using outside solutions is malpractice. |
| Five submissions a day | Unused ones do not carry over. |
Five milestones, each a form on the course portal. You must submit at least two, and one only counts once the response copy reaches your email. In May 2026 they fell on 21 June, 28 June, 5 July, 12 July and 21 July. The competition closed on 29 July, the cutoff was a leaderboard score below 0.20, and both vivas had to be done by 23 August.
Final = NB + M + L + S + V (out of 10 + 5 + 30 + 5 + 50 = 100)
NB your notebook, marked by the examiner. M all milestones done. L your leaderboard position. S how many submissions you made. V the Level 2 viva.
You pass if your viva is at least 25/50 and your final mark is at least 50/100. A viva of 15–24 may still pass with a D or E at the instructor's discretion. Below 15, you repeat the project next term.
Source: MLP Project Guideline, May 2026. Jan and May 2026 MLP students were marked out of 105: up to 5 bonus marks by viva batch (Batch 1: +5 and a reattempt, Batch 2: +2 and a reattempt, Batch 3: none).
The bonus goes to the earlier viva batches, so being ready early pays. Check your own term's guideline for how batches are assigned.
Level 1 is a checklist, not a score. The examiner checks that you understood the problem and loaded the data right; that your EDA looks at single features and pairs of features with readable plots; that missing values, scaling and encoding are handled with reusable pipelines applied the same way to train and test; that you engineered features; that you tried at least three models from different families, tuned them and compared them with suitable metrics; and that the notebook is clean and holds your best model.
Level 2 decides your marks. Concepts, maths, practice and programming. You may be asked to walk through your notebook in five minutes, to explain the five components of any model you used or learned, or to write code on the spot. The examiner can ask about anything in the diploma, MLF, MLT and statistics included. In the guideline's words: "Only using an API (e.g. DecisionTreeClassifier, SVC, etc.) is not sufficient, you should also know how it works internally."
No code, no solution. The competition is different every term.
submission.csv early. Long notebooks time out. Write the file after each stage, so a timeout still leaves a valid submission.Weekly assignments carry little weight, but Weeks 1 to 7 decide whether you may sit the end term at all. Treat the early weeks as the ones that matter most.
Every week of every course has one, with a fixed deadline in the grading document; the May 2026 document says, in capitals, that none of them will move. They are 5% of MLF and MLT and 10% of MLP.
In MLF and MLT you get multiple choice, multiple select and numeric answers. Treat each option of a multiple-select question as its own true-or-false question. In a numeric one, read the rounding instruction first, keep full precision through the working, and round once, at the end. When a definition has two versions (dividing a variance by n or by n−1), use the one the lectures use.
In MLP most questions hand you a dataset and a precise recipe and ask for a number: train this model with these settings, report this metric on this split, to three decimals. The marks go to the exact number, so the job is matching the recipe, not modelling well:
random_state.Type your Week 1–7 scores and Mock Quiz 1. It takes the best 5 of the 8. Leave a box empty if that assignment isn't out yet; it counts as zero. Nothing leaves your browser.
Written in person at an exam centre. Quiz 1 covers Weeks 1–4 and Quiz 2 covers Weeks 1–8. It is one sitting for all your quiz courses: 2 hours for one or two courses, 3 for three, 4 for four. There is no make-up quiz, and a missed quiz scores zero (Academics page). In the Sep 2026 term, Quiz 1 is on Sunday 15 November and Quiz 2 on Saturday 5 December.
Mock Quiz 1 counts: it is in both GAA and the eligibility check.
Each course runs a mock before Quiz 1, and it counts. Where a course runs mocks before both quizzes, an average of 40 or more across them can earn up to 2 bonus marks. Past papers are the best preparation there is; chapter 10 lists students who have solved them on video.
Online, remote-proctored programming exams, on your own computer. Pass the System Compatibility Test first, or OPPE 1 is not scheduled at all. In the Sep 2026 term OPPE 1 runs on 21–22 November, and OPPE 2 on 19–20 December and 2–3 January; your own slot is allotted closer to the date.
The practice that transfers best: redo the weekly MLP assignments against a clock until loading, splitting, fitting and scoring is muscle memory. A student-made OPPE 1 mock on Kaggle shows you the format. Read the exam's own instructions for what may be open on your screen; don't assume.
In person; in the Sep 2026 term, Sunday 10 January 2027. It covers all twelve weeks, including Weeks 11 and 12. It is 40–60% of MLF and MLT depending on which half of the formula wins, and 30% of MLP. In MLF and MLT a strong end term can rescue a weak quiz. In MLP no end term score can stand in for passing an OPPE.
Pick a course, type what you know, and see your total, your grade and the end term score each grade needs.
It uses the exact formulas from chapters 04 to 07 and draws your total against the one score you don't know yet. Numbers stay in this browser.
| Grade | Total needed | End term needed |
|---|
You also need end term eligibility (chapter 04), for MLP an OPPE score of 40 or more (chapter 06), and for the project a viva of 25 or more and a total of 50 (chapter 07). The calculator checks these too.
I opened every link here before adding it. Official material first, because the exams are written from it. Everything else explains the same ideas a second way.
Left out on purpose: repos and videos that post assignment or OPPE answers, and paid "solutions" sites.
They are easy to find and they cost more than they give: the assignments decide your end term eligibility, and the project rules treat outside solutions as malpractice.
Yes. MLF is a co-requisite of MLT, and most plans pair them (chapter 02). Expect MLT to use eigenvectors before MLF teaches them.
The grading document lists MLT as a prerequisite of MLP, so plan MLP for the term after MLT. If your registration page lets you do otherwise, ask the programme office before relying on it.
No to both. MLF and MLT have two in-person quizzes and an end term. MLP has two OPPEs, three Kaggle assignments and an end term.
It scores zero and there is no make-up. The formula takes the better of two versions, and one of them uses only your better quiz, so one missed quiz costs less than you fear. But you must attend at least one quiz to sit the end term.
Yes. Mock Quiz 1 is the "Mock Assignment" on your dashboard, and it is one of the scores in both the end term eligibility check and GAA.
Not towards GAA, which uses Weeks 1–10 and the mock. Their content is in the end term, and the grading document asks you to do them anyway.
The System Compatibility Test checks your computer can run the proctored OPPE. For MLP you need it: without it OPPE 1 is not scheduled for you. Do it the week it opens.
Not necessarily. If you sat the end term and your total is high enough, you get I_OP and take only the OPPEs next term, with your other marks carried over. See the table in chapter 06.
No. The three Kaggle assignments are 20% of the MLP course. The MLP Project is a separate two-credit course with its own competition, milestones and vivas.
No. The guideline requires a CPU-only notebook; don't switch on Kaggle's accelerators.
No. Only the listed libraries: NumPy, pandas, Matplotlib, scikit-learn, XGBoost, Seaborn, imbalanced-learn, SciPy, regex, LightGBM, Plotly, statsmodels, WordCloud, CatBoost and Python's built-ins.
You can upload it, but it won't count. Only submissions from your registered FinalNotebook count, and experiments run on your own machine are not considered for evaluation.
Five a day, and unused ones do not carry over. How many you make is itself worth up to 5 marks.
It is set every term. In May 2026 it was a leaderboard score below 0.20, and your FinalNotebook, when run, had to reach it too.
No. The project and both vivas must be done in the same term; the problem changes every term.
Anything from the diploma, MLF, MLT and statistics included; the five components of every model you used or learned; a walk-through of your notebook; possibly code written on the spot. See chapter 07.
No. Bishop, ESL, ISL, Mathematics for Machine Learning and Boyd are all free from their authors. Only Duda, Hart and Stork is not.
Less than MLT to pass the assignments, but the viva and the Kaggle work reward knowing what each model does inside. That is MLT's job; MLP assumes it.
Don't. Assignments are individual work, and the project guideline calls sharing notebooks or snippets malpractice. Discussing ideas is fine; this guide stays on that side of the line.
On Discourse, where the course teams answer, and in the live and TA sessions (chapter 10).
The usual scale is S 90, A 80, B 70, C 60, D 50, E 40 and above, U below 40; confirm it in your term's handbook. The calculator shows the end term score each grade needs.
Registration is on 22–23 September 2026, the term starts on 2 October, and the end term is on 10 January 2027. All the dates are in the term chart.
No doubt matches that. Try a shorter word, or ask on Discourse.
If you remember nothing else
Attend at least one quiz. The formula forgives one bad quiz; it does not forgive a bad end term.
60% is code. No grade without an OPPE of 40. Take the SCT early.
Register exactly as the guideline says, submit only from the FinalNotebook, keep it private, finish early.