Tagline: Data science and ML fundamentals — maths by hand, code implementations, honest library defaults, supervised pipelines.
| Field | Value |
|---|---|
| Course code | DS-ML-A |
| Duration | 22 weeks |
| Credits | 4 |
| Compliance | NBA-aligned OBE |
| Prerequisite | PY-FUND-A or equivalent Python |
| Publication | published |
| Price | INR 10000 |
| DB snapshot | 2026-10-08T15:04:28.919Z |
Course objective
Compute core DS/ML maths by hand, implement the same ideas in code, use library APIs with honest defaults, and run supervised pipelines on small in-memory course datasets.
Course outcomes (CO1–CO6)
| CO | Bloom | Weight | Statement |
|---|---|---|---|
| CO1 | apply | 15% | Compute descriptive stats, vectors, distances, and losses by hand and in code. |
| CO2 | analyze | 15% | Explain train/test splits, leakage, and baselines; compute classification and regression metrics. |
| CO3 | apply | 14% | Forward-propagate a tiny network, compare activations, and perform backprop by hand and in code. |
| CO4 | apply | 14% | Fit and evaluate linear, logistic, k-NN, tree, Naive Bayes, and linear SVM models with documented library limits. |
| CO5 | evaluate | 14% | Run load → EDA → clean → scale/PCA → model select → metrics on course datasets. |
| CO6 | create | 14% | Author small ML-shaped classes with fit, predict, and transform that mirror library APIs. |
| CO7 | apply | 14% | (Portfolio) Apply the same pipeline to a large CSV locally — instructional only, not platform-graded. |
Activity totals (live from platform)
| Metric | Count |
|---|---|
| Units / chapters | 17 |
| Topics | 206 |
| Topic quiz activities | 1115 |
| MCQ questions | 1115 |
| Practice labs | 1 |
| Evaluated coding labs | 250 |
| Instructional content blocks | 1289 |
Certificates & earning criteria
Certificate rules defined in curriculum — contact support for latest eligibility.
Unit syllabus (unit → topics → activities)
Unit 1 — Orientation & DS/ML map
Unit totals: 5 topics · 27 MCQ questions · 1 practice · 1 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | Course learning roadmap | 7 | 0 | 0 | — |
| 2 | DS/ML roles and workflow map | 5 | 0 | 0 | — |
| 3 | Tabular data shape & features | 6 | 0 | 0 | — |
| 4 | python_applied lab contract (no uploads) | 5 | 1 | 1 | — |
| 5 | Models as objects — usage note | 4 | 0 | 0 | — |
Unit 2 — Descriptive statistics: center, spread & outliers
Unit totals: 13 topics · 68 MCQ questions · 0 practice · 12 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | Why stats: EDA in the ML pipeline | 6 | 0 | 0 | — |
| 2 | What a column is (math before mean) | 6 | 0 | 1 | — |
| 3 | Humans vs typical jobs on this desk | 6 | 0 | 1 | — |
| 4 | English → mean on one list | 6 | 0 | 1 | — |
| 5 | Sepal list shape in code | 6 | 0 | 1 | — |
| 6 | Mean, median, mode (maths) | 5 | 0 | 1 | — |
| 7 | Summaries in code | 4 | 0 | 1 | — |
| 8 | Variance & ddof (maths) | 6 | 0 | 1 | — |
| 9 | Spread metrics in code | 5 | 0 | 1 | — |
| 10 | Range & percentiles (maths) | 5 | 0 | 1 | — |
| 11 | Percentiles in code | 4 | 0 | 1 | — |
| 12 | IQR & outlier rules (maths) | 5 | 0 | 1 | — |
| 13 | Outlier handling in code | 4 | 0 | 1 | — |
Unit 3 — Vectors, distance & scaling for ML
Unit totals: 12 topics · 66 MCQ questions · 0 practice · 12 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | What a vector is (math before flowers) | 6 | 0 | 1 | — |
| 2 | Humans vs computer on these lists | 6 | 0 | 1 | — |
| 3 | One flower as two numbers | 6 | 0 | 1 | — |
| 4 | One vector in code (shape only) | 6 | 0 | 1 | — |
| 5 | Vectors, notation & dot product (maths) | 6 | 0 | 1 | — |
| 6 | NumPy vectors & dot product | 5 | 0 | 1 | — |
| 7 | L1, L2 norms & unit vectors (maths) | 6 | 0 | 1 | — |
| 8 | Distance functions in code | 5 | 0 | 1 | — |
| 9 | z-score standardization (maths) | 5 | 0 | 1 | — |
| 10 | StandardScaler pattern | 5 | 0 | 1 | — |
| 11 | Min–max scaling (maths) | 5 | 0 | 1 | — |
| 12 | MinMaxScaler + scale-before-distance | 5 | 0 | 1 | — |
Unit 4 — Probability, loss & gradients
Unit totals: 16 topics · 75 MCQ questions · 0 practice · 16 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | What a chance is | 6 | 0 | 1 | — |
| 2 | Humans vs computer on these chances | 5 | 0 | 1 | — |
| 3 | Eight flowers, equally likely | 5 | 0 | 1 | — |
| 4 | Axiom rules on chance (code map) | 5 | 0 | 1 | — |
| 5 | Probability axioms & simple events (maths) | 5 | 0 | 1 | — |
| 6 | Simulation & frequency in code | 4 | 0 | 1 | — |
| 7 | Conditional probability & independence (maths) | 5 | 0 | 1 | — |
| 8 | Conditional from counts in code | 4 | 0 | 1 | — |
| 9 | Bayes theorem prior→posterior (maths) | 5 | 0 | 1 | — |
| 10 | Bayes calculator in code | 4 | 0 | 1 | — |
| 11 | MAE, MSE & outlier sensitivity (maths) | 5 | 0 | 1 | — |
| 12 | Loss functions in code | 4 | 0 | 1 | — |
| 13 | Slope & finite differences 1D (maths) | 4 | 0 | 1 | — |
| 14 | Finite-difference slope in code | 4 | 0 | 1 | — |
| 15 | One gradient step + GD/SGD honesty (maths) | 5 | 0 | 1 | — |
| 16 | One GD update in code | 5 | 0 | 1 | — |
Unit 5 — pandas, tables & exploratory data analysis
Unit totals: 12 topics · 64 MCQ questions · 0 practice · 12 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | What a table is | 6 | 0 | 1 | — |
| 2 | Humans vs computer on this table | 6 | 0 | 1 | — |
| 3 | Add the sepal column | 6 | 0 | 1 | — |
| 4 | Frame shape in code | 6 | 0 | 1 | — |
| 5 | Series, DataFrame & column semantics | 5 | 0 | 1 | — |
| 6 | Select, filter, assign in pandas | 5 | 0 | 1 | — |
| 7 | dtypes, missingness counts | 5 | 0 | 1 | — |
| 8 | Impute / cast dtypes on course tables | 5 | 0 | 1 | — |
| 9 | Group-by aggregates & pivot intuition | 5 | 0 | 1 | — |
| 10 | groupby / pivot_table in code | 5 | 0 | 1 | — |
| 11 | Pearson correlation & covariance lite | 5 | 0 | 1 | — |
| 12 | EDA numeric report (describe, corr) | 5 | 0 | 1 | — |
Unit 6 — Evaluation & metrics
Unit totals: 12 topics · 65 MCQ questions · 0 practice · 12 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | What a score pile is (math before splits) | 6 | 0 | 1 | — |
| 2 | Humans vs computer on these scores | 6 | 0 | 1 | — |
| 3 | English → floor split | 6 | 0 | 1 | — |
| 4 | Split counts in code | 6 | 0 | 1 | — |
| 5 | Train/validation/test purpose | 5 | 0 | 1 | — |
| 6 | train_test_split & reproducible seed | 5 | 0 | 1 | — |
| 7 | Confusion matrix arithmetic | 6 | 0 | 1 | — |
| 8 | CM from predictions in code | 5 | 0 | 1 | — |
| 9 | Precision, recall, F1 by hand | 5 | 0 | 1 | — |
| 10 | Classification metrics in code | 5 | 0 | 1 | — |
| 11 | RMSE, MAE, R² lite | 5 | 0 | 1 | — |
| 12 | Regression metrics in code | 5 | 0 | 1 | — |
Unit 7 — Supervised I: Linear & logistic
Unit totals: 13 topics · 68 MCQ questions · 0 practice · 18 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | What a line is (math before SSE) | 6 | 0 | 1 | — |
| 2 | Humans vs miss jobs on this desk | 6 | 0 | 1 | — |
| 3 | English → ŷ on one flower | 6 | 0 | 1 | — |
| 4 | Sepal/petal columns shape in code | 6 | 0 | 1 | — |
| 5 | Line model & least-squares intuition | 5 | 0 | 1 | — |
| 6 | Predict from slope/intercept in code | 5 | 0 | 1 | — |
| 7 | Fit linear model on small table | 5 | 0 | 1 | — |
| 8 | LinearRegression fit/predict | 5 | 0 | 1 | — |
| 9 | Sigmoid & log-odds | 5 | 0 | 1 | — |
| 10 | Logistic curve & threshold in code | 5 | 0 | 1 | — |
| 11 | L2 regularization intuition | 5 | 0 | 1 | — |
| 12 | Ridge / LogisticRegression callout | 5 | 0 | 1 | — |
| 13 | Scenario playground | 4 | 0 | 6 | — |
Unit 8 — Supervised II: k-NN & instance-based learning
Unit totals: 13 topics · 76 MCQ questions · 0 practice · 18 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | What a neighbor is (math before vote) | 6 | 0 | 1 | — |
| 2 | Humans vs memory jobs on this desk | 6 | 0 | 1 | — |
| 3 | English → distance on one pair | 6 | 0 | 1 | — |
| 4 | Memory-board shape in code | 6 | 0 | 1 | — |
| 5 | k-NN vote & tie rules | 6 | 0 | 1 | — |
| 6 | Manual k-NN on tiny table | 6 | 0 | 1 | — |
| 7 | k-NN regression (mean of neighbors) | 6 | 0 | 1 | — |
| 8 | KNeighborsClassifier | 6 | 0 | 1 | — |
| 9 | Weighted neighbors & metric choice | 6 | 0 | 1 | — |
| 10 | KNeighborsRegressor + weights | 6 | 0 | 1 | — |
| 11 | Curse of dimensionality (concept + numeric) | 6 | 0 | 1 | — |
| 12 | Scale-then-k-NN pipeline step | 6 | 0 | 1 | — |
| 13 | Scenario playground | 4 | 0 | 6 | — |
Unit 9 — Supervised III: Trees & ensembles
Unit totals: 13 topics · 76 MCQ questions · 0 practice · 18 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | What a fork is (math before Gini) | 6 | 0 | 1 | — |
| 2 | Humans vs mix jobs on this desk | 6 | 0 | 1 | — |
| 3 | Label bag — pure vs mixed by eye | 6 | 0 | 1 | — |
| 4 | Label column shape in code | 6 | 0 | 1 | — |
| 5 | Entropy & Gini impurity | 6 | 0 | 1 | — |
| 6 | Impurity calculator in code | 6 | 0 | 1 | — |
| 7 | Best split & tree path trace | 6 | 0 | 1 | — |
| 8 | DecisionTreeClassifier / Regressor | 6 | 0 | 1 | — |
| 9 | Bagging & bootstrap intuition | 6 | 0 | 1 | — |
| 10 | RandomForestClassifier | 6 | 0 | 1 | — |
| 11 | Boosting residual intuition | 6 | 0 | 1 | — |
| 12 | GradientBoostingClassifier | 6 | 0 | 1 | — |
| 13 | Scenario playground | 4 | 0 | 6 | — |
Unit 10 — Supervised IV: Naive Bayes, SVM & families
Unit totals: 13 topics · 78 MCQ questions · 0 practice · 18 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | What a family is (math before scores) | 6 | 0 | 1 | — |
| 2 | Humans vs family jobs on this desk | 6 | 0 | 1 | — |
| 3 | English → both stamps on one flower | 6 | 0 | 1 | — |
| 4 | X/y shape for family desks | 6 | 0 | 1 | — |
| 5 | Generative vs discriminative | 6 | 0 | 1 | — |
| 6 | Same toy computed two ways | 6 | 0 | 1 | — |
| 7 | Naive Bayes (Gaussian + categorical lite) | 6 | 0 | 1 | — |
| 8 | GaussianNB / CategoricalNB | 6 | 0 | 1 | — |
| 9 | Max-margin & support vectors intuition | 6 | 0 | 1 | — |
| 10 | LinearSVC / SVC(kernel='linear') | 6 | 0 | 1 | — |
| 11 | Kernel trick concept | 6 | 0 | 1 | — |
| 12 | Model family comparison on same split | 6 | 0 | 1 | — |
| 13 | Scenario playground | 6 | 0 | 6 | — |
Unit 11 — Unsupervised I: Clustering
Unit totals: 13 topics · 78 MCQ questions · 0 practice · 18 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | What a pile is (math before k-means) | 6 | 0 | 1 | — |
| 2 | Humans vs pile jobs on this desk | 6 | 0 | 1 | — |
| 3 | Unlabeled desk — piles by eye | 6 | 0 | 1 | — |
| 4 | Flower matrix shape in code | 6 | 0 | 1 | — |
| 5 | Unsupervised vs supervised recap | 6 | 0 | 1 | — |
| 6 | Load unlabeled & visualize numbers | 6 | 0 | 1 | — |
| 7 | k-means objective & assignment | 6 | 0 | 1 | — |
| 8 | k-means iteration in code | 6 | 0 | 1 | — |
| 9 | Elbow & silhouette numbers | 6 | 0 | 1 | — |
| 10 | KMeans + inertia / silhouette | 6 | 0 | 1 | — |
| 11 | Agglomerative linkage (single/complete lite) | 6 | 0 | 1 | — |
| 12 | AgglomerativeClustering | 6 | 0 | 1 | — |
| 13 | Scenario playground | 6 | 0 | 6 | — |
Unit 12 — Unsupervised II: PCA & representation
Unit totals: 13 topics · 75 MCQ questions · 0 practice · 18 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | What a cloud is (math before variance) | 6 | 0 | 1 | — |
| 2 | Humans vs cloud jobs on this desk | 5 | 0 | 1 | — |
| 3 | Feature point on the PCA plane | 5 | 0 | 1 | — |
| 4 | Flower matrix shape in code | 5 | 0 | 1 | — |
| 5 | Variance as information | 6 | 0 | 1 | — |
| 6 | Column variances in code | 6 | 0 | 1 | — |
| 7 | PCA projection 2D (given v) | 6 | 0 | 1 | — |
| 8 | Manual PCA step on a tiny matrix | 6 | 0 | 1 | — |
| 9 | Explained variance ratio | 6 | 0 | 1 | — |
| 10 | PCA fit/transform | 6 | 0 | 1 | — |
| 11 | PCA for visualization vs modelling | 6 | 0 | 1 | — |
| 12 | PCA → classifier pipeline | 6 | 0 | 1 | — |
| 13 | Scenario playground | 6 | 0 | 6 | — |
Unit 13 — Neural networks I: neuron and forward
Unit totals: 13 topics · 78 MCQ questions · 0 practice · 18 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | Living cell vs artificial unit | 6 | 0 | 1 | — |
| 2 | Humans vs computers on this desk | 6 | 0 | 1 | — |
| 3 | Two functions in one unit (sum then bend) | 6 | 0 | 1 | — |
| 4 | One node in code (Σ then Φ) | 6 | 0 | 1 | — |
| 5 | One neuron is logistic’s cousin | 6 | 0 | 1 | — |
| 6 | Neuron forward in code | 6 | 0 | 1 | — |
| 7 | Stack to NET221: full forward tables on F4 | 6 | 0 | 1 | — |
| 8 | Tiny-net forward, given w | 6 | 0 | 1 | — |
| 9 | Activations compare + failure modes | 6 | 0 | 1 | — |
| 10 | Activation functions in code | 6 | 0 | 1 | — |
| 11 | Why depth: two detectors on the iris plane | 6 | 0 | 1 | — |
| 12 | Vectorized forward vs loop honesty | 6 | 0 | 1 | — |
| 13 | Scenario playground | 6 | 0 | 6 | — |
Unit 14 — Neural networks II: backprop and the wall
Unit totals: 13 topics · 76 MCQ questions · 0 practice · 18 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | Miss-word map (guess, score, update) | 6 | 0 | 1 | — |
| 2 | Humans vs computers scoring a miss | 6 | 0 | 1 | — |
| 3 | English→BCE on F4 | 6 | 0 | 1 | — |
| 4 | Scratchpad: which weight board we may touch | 6 | 0 | 1 | — |
| 5 | Loss on NET221 (BCE) | 6 | 0 | 1 | — |
| 6 | Same loss in code | 6 | 0 | 1 | — |
| 7 | Chain rule, one step (W_O[0] only) | 6 | 0 | 1 | — |
| 8 | One backprop step in code | 6 | 0 | 1 | — |
| 9 | Epoch loss table + GD/SGD names | 6 | 0 | 1 | — |
| 10 | Tiny loop vs MLPClassifier | 6 | 0 | 1 | — |
| 11 | The wall: flatten vs 4-vector | 6 | 0 | 1 | — |
| 12 | Library MLP + named limits | 6 | 0 | 1 | — |
| 13 | Scenario playground | 4 | 0 | 6 | — |
Unit 15 — OOP-for-ML
Unit totals: 13 topics · 59 MCQ questions · 0 practice · 18 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | Object-word map (class/instance/self/state) | 5 | 0 | 1 | — |
| 2 | Humans vs computers: sticky note vs object that remembers | 5 | 0 | 1 | — |
| 3 | English→RunningStats after ZOOM4 (n=4,sum=22,mean=5.5) | 5 | 0 | 1 | — |
| 4 | Scratchpad: attribute box shape only | 4 | 0 | 1 | — |
| 5 | Class, instance, self (engine) | 5 | 0 | 1 | — |
| 6 | Author RunningStats | 4 | 0 | 1 | — |
| 7 | fit stores learned attributes | 5 | 0 | 1 | — |
| 8 | Author DotProductRegressor | 4 | 0 | 1 | — |
| 9 | predict & fit-before-predict | 5 | 0 | 1 | — |
| 10 | Author ThresholdClassifier | 4 | 0 | 1 | — |
| 11 | transform & train/test leakage | 5 | 0 | 1 | — |
| 12 | Compose scaler → model by hand | 4 | 0 | 1 | — |
| 13 | Playground 1…6 | 4 | 0 | 6 | — |
Unit 16 — Pipelines & tuning
Unit totals: 13 topics · 59 MCQ questions · 0 practice · 18 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | Leak-word map (train/test/fit_stats/score) | 5 | 0 | 1 | — |
| 2 | Humans vs computers: honest split vs peeking | 5 | 0 | 1 | — |
| 3 | English→k-fold on n=8 k=4 (fold_size=2, test_once=1) | 5 | 0 | 1 | — |
| 4 | Scratchpad: pipeline step slots shape only | 4 | 0 | 1 | — |
| 5 | Data leakage patterns (train vs leak centering) | 5 | 0 | 1 | — |
| 6 | ColumnTransformer / encoders | 4 | 0 | 1 | — |
| 7 | k-fold CV arithmetic + CV mean 0.75 | 5 | 0 | 1 | — |
| 8 | cross_val_score | 4 | 0 | 1 | — |
| 9 | Bias–variance & grid search concept | 5 | 0 | 1 | — |
| 10 | GridSearchCV lite | 4 | 0 | 1 | — |
| 11 | Pick metric for business goal | 5 | 0 | 1 | — |
| 12 | Model selection report dict | 4 | 0 | 1 | — |
| 13 | Playground 1…6 | 4 | 0 | 6 | — |
Unit 17 — Capstones
Unit totals: 6 topics · 27 MCQ questions · 0 practice · 5 evaluated labs
| # | Topic | MCQs | Practice | Evaluated | Preview |
|---|---|---|---|---|---|
| 1 | Capstone brief & rubric | 5 | 0 | 1 | — |
| 2 | Shared prep: clean + scale + split | 4 | 0 | 1 | — |
| 3 | Capstone A — Classification | 5 | 0 | 1 | — |
| 4 | Capstone B — Regression | 5 | 0 | 1 | — |
| 5 | Optional: unsupervised EDA add-on | 4 | 0 | 1 | — |
| 6 | Local portfolio appendix | 4 | 0 | 0 | — |
Counts are from the live course catalog. MCQs = questions in topic checkpoint quizzes. Practice = formative run-only or SQL practice lane. Evaluated = submit-graded labs that count toward progress and certificates.