Source document. This week-by-week plan was written for the original six-semester programme and is kept as a reference. The courses on this site can be studied in any order; take the schedule as one way through them, not the only one.
A week-by-week schedule for the whole programme — Semesters I to VI, and both
elective tracks — weighted by the difficulty established in
SYLLABUS-REVIEW.md.
You take fifteen courses: Courses 1–11, then one track's pair in Semester V and the same track's pair in Semester VI. All nineteen course numbers are scheduled below so you can see both tracks before choosing.
The weighting is the point. An even split across five units is the wrong plan when one unit holds three units' worth of material. Where a unit gets more weeks than its neighbours below, the review explains why.
Two courses. The lightest semester of the degree; use the slack to build programming habits that will carry you through the next two years.
| Week | Course 1 — Computer Fundamentals | Course 2 — Problem Solving Using C |
|---|---|---|
| 1 | Unit 1 — number systems: binary, decimal | Unit 1 — software types, compiler vs interpreter |
| 2 | Unit 1 — octal, hex, conversions | Unit 1 — algorithms, flowcharts, C history |
| 3 | Unit 1 — binary arithmetic, complements, generations | Unit 1 — tokens, data types, operators, I/O |
| 4 | Unit 2 — memory hierarchy, storage | Unit 2 — if, if-else, else-if ladder |
| 5 | Unit 2 — types of computers, networks | Unit 2 — switch, loops |
| 6 | Unit 2 — topologies, Internet basics | Unit 2 — break, continue, goto; patterns |
| 7 | Revision 1 — Units 1–2 both courses | Revision 1 |
| 8 | Unit 3 — Word: formatting, styles, tables | Unit 3 — 1-D arrays, memory representation |
| 9 | Unit 3 — mail merge, presentations, shortcuts | Unit 3 — 2-D arrays, matrix operations |
| 10 | Unit 4 — cell referencing, basic functions | Unit 3 — strings, string functions |
| 11 | Unit 4 — logical and text functions | Unit 4 — pointers ⚠ |
| 12 | Unit 4 — VLOOKUP, XLOOKUP, INDEX+MATCH | Unit 4 — functions, recursion, parameter passing ⚠ |
| 13 | Unit 5 — pivot tables, slicers | Unit 4 — storage classes ⚠ |
| 14 | Unit 5 — what-if analysis, dashboards | Unit 5 — dynamic memory, structures, unions |
| 15 | Revision 2 — full syllabus | Unit 5 — file handling; Revision 2 |
Course 2 Unit 4 gets three weeks (11–13). It is the hardest material in the first year, and the syllabus mislabels it "Functions" when it opens with pointers — see finding D5. Draw memory diagrams for every pointer example.
Two courses. The most important semester in the degree. Python and statistics are the foundation of everything from Semester III onward.
| Week | Course 3 — Python | Course 4 — Statistics |
|---|---|---|
| 1 | Unit 1 — features, modes, identifiers, types | Unit 1 — probability rules, axioms |
| 2 | Unit 1 — operators, precedence, I/O | Unit 1 — conditional probability, Bayes ⚠ |
| 3 | Unit 2 — control flow, for…else |
Unit 1 — central tendency, dispersion |
| 4 | Unit 2 — functions, arguments, scope, lambda | Unit 2 — random variables, PMF/PDF/CDF |
| 5 | Unit 3 — strings and lists | Unit 2 — expectation, variance, moments |
| 6 | Unit 3 — tuples, sets | Unit 3 — binomial, Poisson |
| 7 | Revision 1 — Units 1–3 | Revision 1; Unit 3 — geometric, negative binomial |
| 8 | Unit 3 — dictionaries, comprehensions | Unit 3 — normal distribution, z-scores |
| 9 | Unit 4 — file handling, CSV ⚠ | Unit 3 — exponential, gamma, CLT |
| 10 | Unit 4 — exception handling ⚠ | Unit 4 — covariance, correlation |
| 11 | Unit 4 — classes, objects, encapsulation ⚠ | Unit 4 — regression, least squares |
| 12 | Unit 4 — inheritance, MRO, polymorphism ⚠ | Unit 4 — residuals, R², ANOVA |
| 13 | Unit 5 — linked lists, stacks, queues | Unit 5 — estimation, confidence intervals |
| 14 | Unit 5 — priority queues; Tkinter | Unit 5 — z-test, t-test |
| 15 | Revision 2 — full syllabus | Unit 5 — chi-square, F-test, errors, power |
Course 3 Unit 4 gets four weeks (9–12), not two. It contains file handling and exception handling and the whole of object-oriented programming — three units compressed into one. Compare Unit 1, which covers only literals and operators. Finding D6.
Course 4 Units 3, 4 and 5 get three weeks each. Distributions, regression and inference are where the marks and the difficulty both are.
Bayes' theorem is scheduled in week 2 even though it is not in the syllabus unit list, because it is examined. Finding D1.
Three courses — half as much again as Semesters I and II, and it lands all at once. Semesters III and IV are the heaviest in the programme. Plan for it before week 1 rather than discovering it in week 6: the two-course rhythm that worked for a year will not survive contact with three.
| Week | Course 5 — DBMS | Course 6 — Data Science with R | Course 7 — Web Technologies |
|---|---|---|---|
| 1 | Unit 1 — data vs information, file-based systems | Unit 1 — the data science process, lifecycle | Unit 1 — HTML structure, elements, attributes |
| 2 | Unit 1 — three-schema architecture, data independence | Unit 1 — EDA, feature engineering | Unit 1 — headings, images, tables, lists |
| 3 | Unit 2 — ER building blocks, entity and attribute types | Unit 2 — R and RStudio, types, operators | Unit 1 — forms and every input type |
| 4 | Unit 2 — relationships, cardinality, participation | Unit 2 — control structures, apply, functions |
Unit 2 — selectors, combinators, the box model |
| 5 | Unit 2 — reducing ER to tables; EER | Unit 2 — packages, I/O (CSV, Excel, JSON) | Unit 2 — position, float, Flexbox and Grid |
| 6 | Unit 3 — relational model, keys, constraints | Unit 3 — data frames, lists, matrices | Unit 2 — pseudo-classes, transitions, CSS forms |
| 7 | Revision 1; Unit 3 — relational algebra | Revision 1; Unit 3 — dplyr and tidyr | Revision 1; Unit 3 — JS basics, variables, operators |
| 8 | Unit 3 — functional dependencies, 1NF, 2NF | Unit 3 — missing data, dates and times | Unit 3 — strings, arrays, functions |
| 9 | Unit 3 — 3NF, BCNF, worked normalization | Unit 3 — ggplot2: grammar, aesthetics, geoms | Unit 3 — objects, regular expressions, exceptions |
| 10 | Unit 4 — DDL, constraints, DML | Unit 3 — faceting, layering, exporting | Unit 4 — the DOM and form elements |
| 11 | Unit 4 — SELECT, aggregates, GROUP BY, HAVING | Unit 4 — simple and multiple regression | Unit 4 — validation, responsive messages |
| 12 | Unit 4 — joins (inner, left, self, three-table) | Unit 4 — accuracy, confusion matrix, ROC | Unit 4 — dialogs, windows, keyboard and mouse events |
| 13 | Unit 4 — set operations, subqueries, views | Unit 4 — K-Means, text mining, recommenders, ethics | Unit 5 — JSON syntax, parsing, nested access |
| 14 | Unit 5 — PL/SQL blocks, cursors, exceptions | Unit 5 — time series: ts, decomposition, ACF/PACF |
Unit 5 — jQuery selectors, DOM manipulation, chaining |
| 15 | Unit 5 — procedures, functions, TRIGGERS ⚠; Revision 2 | Unit 5 — ARIMA, forecasting; plotly and Shiny | Unit 5 — events, animations; Revision 2 |
Triggers are scheduled in week 15 despite being absent from the Course 5 Unit 5 syllabus list. Two of the six PL/SQL lab questions are trigger problems. Finding D2.
Course 5 weeks 8–9 (normalization) and week 12 (joins) carry the most exam weight in that course. Nothing else in DBMS is asked as reliably.
Course 6 Unit 5 gets two weeks (14–15) and is still tight. It fuses three unrelated subjects — ARIMA time series, plotly interactivity, and building Shiny web applications — any one of which would be a unit elsewhere. If you run short, ARIMA is the examinable half; Shiny is the one that gets a descriptive question rather than a technical one.
Course 7 is the lightest of the three to learn and the heaviest to practise. Nothing in it is conceptually hard, and its 16 lab experiments still take longer than the other two courses' labs combined, because each one has to be built and looked at. Do the lab work in the week the topic is taught, not in a block before the exam.
They are taught in the same semester and they meet in Semester IV. Course 5's
SELECT … WHERE … GROUP BY becomes Course 10's aggregation pipeline; Course
7's JSON becomes Course 10's document. Every hour you spend understanding
why a join exists in DBMS pays out twice more before the degree ends.
Two weeks, during the break before Semester IV begins.
Regression, correlation and hypothesis testing are taught in Semester II. Their first real application is Data Mining in Semester IV, and Machine Learning is a Year III elective in Semester V — three semesters after the theory. Statistical intuition decays without use. Finding D10.
| Day | Revise |
|---|---|
| 1–2 | Descriptive statistics, distributions (formula sheet) |
| 3–4 | Correlation and regression; re-run 04_correlation_regression.py |
| 5–6 | Hypothesis testing; re-run 05_inference_hypothesis_tests.py |
| 7–10 | Bridge the Excel/Python gap — redo the stats labs in Python (finding D8) |
| 11–14 | Python revision: NumPy and Pandas basics, ready for Course 9 |
That last block matters. The Semester II stats lab is entirely Excel, so you arrive in Semester IV able to run a regression in a spreadsheet but not in the language you spent a semester learning.
Three courses again — and this is the semester the degree has been building towards. Course 9 is the one you will use in every job you take; Course 8 is where the Semester II statistics finally gets applied; Course 10 is Course 5 seen from the other side.
| Week | Course 8 — Data Mining | Course 9 — Python for Data Analysis | Course 10 — Document Database |
|---|---|---|---|
| 1 | Unit 1 — warehouse vs database, characteristics | Unit 1 — the ndarray, dtypes, creating arrays | Unit 1 — NoSQL, its history and features |
| 2 | Unit 1 — architecture, star and snowflake schemas | Unit 1 — indexing, slicing, views vs copies | Unit 1 — CAP and BASE against ACID; the four types |
| 3 | Unit 1 — fact constellation, OLAP cube operations | Unit 1 — broadcasting, ufuncs, axis, random |
Unit 1 — RDBMS vs NoSQL; JSON and BSON; install |
| 4 | Unit 2 — KDD vs data mining, tasks | Unit 2 — Series, DataFrame, Index objects | Unit 2 — database/collection/document, BSON format |
| 5 | Unit 2 — cleaning, missing data, dimensionality reduction | Unit 2 — loc vs iloc, boolean filtering |
Unit 2 — data types, ObjectId, the type traps |
| 6 | Unit 2 — discretization, transformation, similarity measures | Unit 2 — alignment, sorting, ranking, duplicates | Unit 2 — schema design; embed or reference |
| 7 | Revision 1; Unit 3 — association rules, support and confidence | Revision 1; Unit 3 — read_csv and the parameters that matter |
Revision 1; Unit 3 — insert, and find with comparison operators |
| 8 | Unit 3 — Apriori, worked by hand | Unit 3 — JSON, json_normalize, Excel |
Unit 3 — logical, element and evaluation operators |
| 9 | Unit 3 — Partition, Pincer-Search, DIC, FP-Growth | Unit 3 — missing data, outliers, map/apply |
Unit 3 — update, replaceOne, delete; arrays |
| 10 | Unit 4 — decision trees, best split, ID3 by hand | Unit 4 — the .str accessor, regex, extract |
Unit 4 — embedded vs normalized, the trade-offs |
| 11 | Unit 4 — C4.5, CART, comparing classifiers | Unit 4 — feature engineering, dummies, sampling | Unit 4 — the three relationships; design patterns |
| 12 | Unit 4 — rule-based, k-NN, Bayesian classifiers | Unit 5 — merge and the four join types | Unit 4 — the aggregation framework, $match/$group |
| 13 | Unit 5 — K-Means to convergence; k-Medoid | Unit 5 — pivot, melt, stack, unstack | Unit 5 — projection, sort, limit, skip; pagination |
| 14 | Unit 5 — hierarchical linkage, DBSCAN, BIRCH | Unit 5 — group-by; recompute the Course 4 examples | Unit 5 — indexes: compound, multikey, text; ESR |
| 15 | Unit 5 — STIRR, ROCK, CACTUS; Revision 2 | Unit 5 — matplotlib, Seaborn, Plotly; Revision 2 | Unit 5 — replication and failover; GridFS ⚠; Revision 2 |
Course 8 is arithmetic, not reading. Apriori, ID3 and K-Means are all
asked as hand traces: given this table, compute the support, the information
gain, the centroids after two iterations. Weeks 8, 10 and 13 are the ones to
protect, and the only way to prepare is to work the traces on paper until the
arithmetic is automatic. The notes give each one fully worked, and
labs/course-8-datamining/ recomputes every figure
in code so you can check your own work against something that runs.
Course 9 Units 1 and 2 get five weeks between them (1–6). Everything after
them assumes the ndarray and the DataFrame, and a student who is still unsure
whether loc is inclusive will lose time in every week that follows. This is
the course that pays out longest — treat it as the priority when three courses
collide.
Course 10 Unit 4 gets three weeks (10–12) because embed-or-reference is the whole subject in one question, and the aggregation pipeline is the other half of the paper.
GridFS is scheduled in week 15 although it appears in no unit's topic list — it survives only in Course Outcome 4 and lab experiment 18. Same shape as findings D1 and D2; here it is finding D13.
The three courses overlap far more than their titles suggest, and using that is the difference between one semester's work and three separate efforts:
| When you meet | In another course you already have |
|---|---|
| Course 8's preprocessing (week 5) | Course 9's cleaning (week 9) — the same operations, in code |
| Course 8's K-Means (week 13) | Course 9's DataFrames; the lab does it in scikit-learn |
Course 9's merge and join types (week 12) |
Course 5's SQL joins — identical semantics, one exception: Pandas joins NaN to NaN and SQL never joins NULL to NULL |
| Course 10's aggregation (week 12) | Course 5's GROUP BY/HAVING, and Course 9's groupby |
| Course 10's documents (week 4) | Course 7's JSON |
| Course 8's classification metrics (week 11) | Course 6's confusion matrix and ROC |
Say the connection out loud in the viva. "This $group is a GROUP BY,
and this second $match is the HAVING" is worth more than a memorised
pipeline, in every one of these three courses.
You choose once, and it binds both Semester V and Semester VI. There is no switching in January.
| Track A — Modelling / AI | Track B — Infrastructure / Engineering | |
|---|---|---|
| Sem V | Machine Learning, Artificial Intelligence | Big Data Technologies, Cloud Computing |
| Sem VI | Neural Networks & Deep Learning, NLP | Time Series Analysis, Data Engineering & MLOps |
| Builds on | Courses 4, 8, 9 | Courses 3, 5, 10 |
| You will spend your time | fitting and evaluating models | moving data and running systems |
| Job titles | Data Scientist, ML Engineer, Research | Data Engineer, Platform Engineer, MLOps |
| Maths load | heavier — probability, linear algebra | lighter; more systems and SQL |
| The honest test | do you enjoy Course 4? | do you enjoy Course 5 and Course 10? |
Both tracks are written out in full in this repository, so read the two Semester V READMEs before deciding. And note the market reality worth knowing: there are more data-engineering jobs than data-science jobs, and Track B graduates are hired faster. That is not a reason to pick it if you dislike the work — but it is a fact worth having.
Three courses again — Course 11 plus your two track courses.
Course 11 is the lightest of the three and the most immediately employable; your two track courses are where the effort goes.
| Week | Topic |
|---|---|
| 1 | Unit 1 — BI concepts, the BI lifecycle, BI against BA |
| 2 | Unit 1 — the data warehouse; star and snowflake; OLAP against OLTP |
| 3 | Unit 2 — Power BI Desktop, connecting and shaping data |
| 4 | Unit 2 — Power Query: step order matters; the M language |
| 5 | Unit 3 — the data model, relationships, cardinality; the fan trap |
| 6 | Unit 3 — DAX: calculated columns against measures; row and filter context |
| 7 | Revision 1; Unit 3 — CALCULATE, and why it is the whole language |
| 8 | Unit 4 — Tableau, dimensions and measures, the shelves |
| 9 | Unit 4 — LOD expressions: FIXED, INCLUDE, EXCLUDE |
| 10 | Unit 5 — dashboard components, filters, slicers, drill-down |
| 11 | Revision 2; Unit 5 — visualization principles and accessibility |
| 12 | Unit 5 — publishing, storytelling, insight communication |
| 13 | Lab: the click-paths, worked end to end on a real dataset |
| 14 | Revision 3 — the whole course |
| 15 | Past papers |
| Week | Course 12 A — Machine Learning | Course 13 A — Artificial Intelligence |
|---|---|---|
| 1 | Unit 1 — types of learning, Mitchell's definition | Unit 1 — definition and scope, history, the Turing Test |
| 2 | Unit 1 — the ML pipeline; types and structure of data | Unit 1 — agents, rationality, PEAS, environments |
| 3 | Unit 2 — preprocessing IN ORDER; split first | Unit 2 — state space, the four components; problem formulation |
| 4 | Unit 2 — why accuracy lies; the confusion matrix | Unit 2 — BFS, DFS, UCS; completeness and optimality |
| 5 | Unit 2 — cross-validation, bias and variance, PCA | Unit 2 — iterative deepening; the Romania map by hand |
| 6 | Unit 3 — simple and multiple regression; LINE | Unit 3 — heuristics, admissibility, consistency |
| 7 | Revision 1; Unit 3 — logistic regression and the odds ratio | Revision 1; Unit 3 — A*, and what an inadmissible heuristic costs |
| 8 | Unit 3 — MLE; Ridge, Lasso, elastic net | Unit 3 — hill climbing, simulated annealing, genetic algorithms |
| 9 | Unit 4 — Naive Bayes, k-NN, decision trees | Unit 3 — CSPs, backtracking, MRV and LCV |
| 10 | Unit 4 — SVM, the margin and the kernel trick; random forest | Unit 4 — propositional logic, truth tables, entailment |
| 11 | Revision 2; Unit 5 — K-Means, WCSS, the elbow | Revision 2; Unit 4 — FOL, unification, forward and backward chaining |
| 12 | Unit 5 — hierarchical, DBSCAN, validation metrics | Unit 4 — resolution and CNF; Unit 5 — expert systems |
| 13 | Unit 5 — the four case studies; lab experiments 1–12 | Unit 5 — Bayes, belief networks, fuzzy logic, ethics |
| 14 | Revision 3 | Revision 3; the 19 Prolog programs, typed and run |
| 15 | Past papers | Past papers |
| Week | Course 12 B — Big Data Technologies | Course 13 B — Cloud Computing |
|---|---|---|
| 1 | Unit 1 — what big data means; the five Vs | Unit 1 — the NIST definition; the five characteristics |
| 2 | Unit 1 — the ecosystem by layer; data locality | Unit 1 — SOA and web services; the four-part architecture |
| 3 | Unit 2 — HDFS: blocks, NameNode, DataNodes | Unit 1 — IaaS, PaaS, SaaS by who manages what |
| 4 | Unit 2 — the block arithmetic and the small-files problem | Unit 2 — virtualization; type 1 and type 2; the six types |
| 5 | Unit 2 — replication, rack awareness, fault tolerance | Unit 2 — overcommit; the four deployment models |
| 6 | Unit 2 — YARN and the three schedulers | Unit 3 — block, file and object storage |
| 7 | Revision 1; Unit 3 — map, shuffle, reduce; word count | Revision 1; Unit 3 — storage classes and the retrieval trap |
| 8 | Unit 3 — combiners, and when they are unsafe; partitioning and skew | Unit 3 — egress and data gravity; key-value databases |
| 9 | Unit 3 — Hive: partitioning, bucketing, managed vs external | Unit 3 — batch vs streaming; the cloud data warehouse |
| 10 | Unit 3 — Pig, Crunch, and the abstraction ladder | Unit 4 — ML in the cloud; AIaaS and GPUaaS |
| 11 | Revision 2; Unit 4 — Sqoop, splits, and the DELETE it never sees | Revision 2; Unit 4 — managed platforms; AutoML and its limits |
| 12 | Unit 4 — Flume, back-pressure; Avro and Parquet | Unit 5 — the four selection factors; the six steps |
| 13 | Unit 5 — HBase, row-key design; ZooKeeper; Spark | Unit 5 — deployment shapes, monitoring, autoscaling, drift |
| 14 | Revision 3; the 17 experiments | Revision 3; the 15 experiments |
| 15 | Past papers | Past papers |
THE BIG IDEA
These are the highest-value hours in the semester, because they catch misunderstandings that a single course cannot.
Course 12 A's regression against Course 4's. The same ten (hours, score) pairs must give slope 4.3030, intercept 43.0303, R² 0.9958. If your hand calculation and scikit-learn disagree, one of them is wrong.
Naive Bayes across Courses 8, 12 A and 13 A. The same play-tennis table must give 0.005291 and 0.020571, normalising to 79.54% no.
The star schema across Courses 11, 12 B and 13 B. South revenue must be ₹10,360 in DAX, in Hive, in Spark and in the cloud warehouse. Four engines, nine rows.
If you can reproduce all three, you have understood the material rather than memorised three separate procedures.
Type and run every lab program. Semester V prescribes 46 experiments on Track A (15 + 12 + 19) and 47 on Track B (15 + 17 + 15) across the three courses you take. Reading them is worth almost nothing; running them, breaking them and fixing them is where the marks are — and it is what the viva examines.
Two courses — the lightest semester on paper, and the one where the project usually lands. You stay in the track you chose for Semester V: 14 A + 15 A or 14 B + 15 B.
NOTE
Within each track the two courses overlap heavily, and doing the overlapping units in the same week roughly halves the work.
| Track | The overlap |
|---|---|
| A | Course 14 A Units 4–5 (RNN, LSTM, attention, transformers) and Course 15 A Units 4–5 are the same architectures, from two directions |
| B | Course 14 B's forecasting and Course 15 B's drift detection are both distribution change over time; PSI and the ADF test answer related questions |
| Weeks | 14 A — Deep Learning | 15 A — NLP |
|---|---|---|
| 1–2 | Unit 1: perceptron, the XOR proof, activations, loss | Unit 1: ambiguity, regex, NLTK and spaCy |
| 3–4 | Unit 2: propagation, initialisation, the learning rate | Unit 2: tokenization, stemming, grammars |
| 5–6 | Unit 2: dropout, batch norm, Keras | Unit 2: parsing, CYK, semantic analysis |
| 7–8 | Unit 3: conv arithmetic, pooling, LeNet/AlexNet/VGG | Unit 3: NER, TF-IDF, n-grams |
| 9–10 | Unit 4: RNN, LSTM, GRU ← | Unit 3–4: embeddings, RNN ← do these together |
| 11–12 | Unit 5: attention, transformers ← | Unit 4–5: transformers, BERT ← and these |
| 13 | Unit 5: transfer learning, AI ethics | Unit 5: summarization, chatbots |
| 14–15 | revision, labs, project | revision, labs, project |
| Weeks | 14 B — Time Series | 15 B — Data Engineering & MLOps |
|---|---|---|
| 1–2 | Unit 1: stationarity, ACF and PACF | Unit 1: the lifecycle, ETL vs ELT |
| 3–4 | Unit 2: ARMA, AIC/BIC, Ljung-Box | Unit 2: architecture, monolith vs microservices |
| 5–6 | Unit 3: ADF and KPSS, ARIMA | Unit 2: batch vs event-driven, TCO |
| 7–8 | Unit 3: SARIMA, prediction intervals | Unit 3: MLflow, reproducibility, DVC |
| 9–10 | Unit 4: VAR, Granger, Kalman | Unit 4: deployment, CI/CD and the metric gate |
| 11–12 | Unit 5: forecast evaluation ← | Unit 5: drift detection ← do these together |
| 13 | Unit 5: Holt-Winters, MASE | Unit 5: GDPR, governance, Responsible AI |
| 14–15 | revision, labs, project | revision, labs, project |
1. The labs are heavier than the timetable suggests. Track A prescribes 26 experiments (12 + 14) and Track B 29 (13 + 16), and several are genuinely slow to run — training a CNN or a transformer is not a two-hour exercise on a laptop.
Start the lab record in week 3, not week 12.
2. Some experiments need tools your lab may not have. This repository records exactly which, and why, for every one — Hugging Face, Kafka, Hadoop, Docker, GitHub Actions, Prometheus. Check what is available in your lab in week 1, and agree with your instructor what the substitute is before you need it.
KEY INSIGHT
Take one dataset through both of your courses. On Track A, classify it and then explain the classification. On Track B, forecast it and then deploy the forecaster with monitoring. A single worked example that crosses both courses is worth more in a viva than two separate ones, and it is the shape of an actual job.
A schedule you can actually keep beats an ambitious one you abandon in week 3.
| Day | Focus |
|---|---|
| Mon–Fri | Attend, then spend 1 hour per subject the same evening consolidating |
| Saturday | 3 hours — lab programs, typed and run, not copied |
| Sunday | 2 hours — revise the week; 1 hour — revise something from three weeks ago |
The Sunday spaced-revision hour is the highest-value hour of the week. Re-reading this week's material feels productive and mostly is not; retrieving three-week-old material is what moves it into long-term memory.
| Cycle | When | What |
|---|---|---|
| Daily | Same evening | Review the day's notes — 15 minutes |
| Weekly | Sunday | The week's units, plus one older topic |
| Revision 1 | Week 7 | Units 1–3 of every course |
| Revision 2 | Week 15 | Full syllabus, past papers |
| Pre-exam | Final fortnight | See below |
| Days | Activity |
|---|---|
| 14–11 | One full pass of every unit's notes |
| 10–8 | Formula sheets and the "mistakes that cost marks" section of each unit |
| 7–5 | Past papers under timed conditions |
| 4–3 | Practice problems; re-work anything you got wrong |
| 2–1 | Formula sheets and quick self-tests only — no new material |
| Exam eve | Sleep. Cramming past midnight costs more than it gains. |
Past papers are the highest-value revision there is. They reveal which topics actually recur, how questions are phrased, and how marks are distributed — none of which the syllabus tells you.
Tick a unit only when you can (a) explain it without notes and (b) solve a problem on it unaided.
Course 1 — Computer Fundamentals
Course 2 — Problem Solving Using C
Course 3 — Python Programming and Data Structures
Course 4 — Statistical Foundations
Course 5 — Database Management Systems
Course 6 — Data Science with R
Course 7 — Web Technologies
Course 8 — Data Mining
Course 9 — Python for Data Analysis and Visualization
Course 10 — Document Oriented Database
mongod or AtlasCourse 11 — Business Intelligence Tools (everyone)
CALCULATECourse 12 A — Machine Learning
Course 13 A — Artificial Intelligence
Course 12 B — Big Data Technologies
Course 13 B — Cloud Computing for Data Science
You stay in the track you chose for Semester V. 14 A pairs with 15 A; 14 B pairs with 15 B.
Course 14 A — Neural Networks and Deep Learning
√d_k, transformers, BERT vs GPT, transfer learning, ethics ⚠ (D29 — the "12 pages" is almost certainly 1-2)Course 15 A — Natural Language Processing
Course 14 B — Time Series Analysis and Forecasting
ct regression, ARIMA, SARIMACourse 15 B — Data Engineering and MLOps
If you do nothing else from this plan:
Type every lab program. Reading code teaches nothing. Everything in
labs/ that can run here does — type them, break them, fix them.
Give Course 3 Unit 4 and Course 4 Units 3–5 double time. They carry the difficulty and the marks.
Study Bayes' theorem and database triggers, though neither appears in its syllabus unit list. Both are examined.
Do past papers under timed conditions, starting a week before the exam.