A data science course syllabus runs in five blocks: Python, SQL, statistics and probability, machine learning, and a capstone project, with deep learning and text analysis usually sitting between the last two. Indian certificate programmes cover that in six to twelve months. A degree covers the same ground plus the mathematics underneath it, and takes three to four years.
This page sets out what belongs in each module, which tools each one actually uses, roughly how many weeks it needs, and what a certificate syllabus deliberately leaves out that a degree does not.
The syllabus at a glance
Very few providers publish a week-by-week split, so the weeks below are how the blocks fall in practice on a programme in the six to twelve month band. Published durations sit across that range: AlmaBetter lists 6 to 8 months and Intellipaat 7 months, upGrad’s IIIT Bangalore executive certificate 6 to 9 months, Coding Ninjas’ data science bootcamp 9 months, and Scaler and Great Learning 12 months each, as listed in September 2026.
| Module | Weeks | Tools you will use | What you should have at the end |
|---|---|---|---|
| Python foundations | 4 to 6 | Python, pandas, NumPy, Jupyter | A notebook that cleans and reshapes a messy dataset unaided |
| SQL | 3 to 4 | MySQL or PostgreSQL | Queries with joins, CTEs and window functions against a real schema |
| Statistics and probability | 4 to 6 | Python, SciPy, statsmodels | A tested hypothesis with its assumptions and limits written down |
| Machine learning, core | 6 to 8 | scikit-learn, matplotlib, seaborn | Two supervised models, evaluated properly and compared |
| Machine learning, advanced | 4 to 6 | scikit-learn, XGBoost | A tuned ensemble model with a written error analysis |
| Deep learning and text analysis | 4 to 6 | TensorFlow or PyTorch, NLTK or spaCy | One neural network and one text project end to end |
| Capstone and portfolio | 4 to 6 | All of the above, plus a BI tool | Three defensible projects and a presentation of each |
Two readings of that table. Machine learning takes the largest share, which is the single difference from an analytics syllabus, where SQL does. And statistics sits before machine learning for a reason: a model you cannot evaluate statistically is a model you cannot defend in an interview.
Module 1: Python foundations, 4 to 6 weeks
Everything downstream is written in Python, so this module decides how hard the rest of the course feels.
- Syntax, control flow, functions, and list, dictionary and set handling
- pandas: reading files, indexing, filtering, groupby, merge, pivot and missing-value handling
- NumPy: arrays, vectorised operations and broadcasting, which is what makes scikit-learn make sense later
- Visual exploration with matplotlib and seaborn
- Working in Jupyter and writing a notebook someone else can follow
A syllabus that spends these weeks on object-oriented design patterns and file handling is teaching general programming. Our Python for data analysis guide covers the slice that matters here.
Module 2: SQL, 3 to 4 weeks
Data science interviews in India still open with SQL, because the data you model has to come out of a warehouse before you can touch it.
- SELECT, filters, GROUP BY and aggregates
- Joins across four or more tables, and recognising a silently dropped row
- Subqueries and common table expressions
- Window functions: ranking, running totals, LAG and LEAD
- Reading a schema you did not design
Our explainer on SQL joins covers the part beginners get wrong most often, and the data science interview questions set shows the level these come up at.
Module 3: Statistics and probability, 4 to 6 weeks
This is the module that separates a data science syllabus from an analytics one, and the one self-taught candidates most often skip.
- Descriptive statistics, distributions and the central limit theorem
- Probability, conditional probability and Bayes theorem
- Sampling, standard error and confidence intervals
- Hypothesis testing, p-values, type I and type II errors, and A/B test design
- Linear and logistic regression as statistical models, before they reappear as machine learning ones
- Enough linear algebra and calculus to follow gradient descent, which is where an analytics syllabus stops and this one continues
If a programme covers statistics in a single week, it is an analytics course with a machine learning module bolted on. That is a legitimate product, and our comparison of data analytics against data science sets out which one you actually need, but it should not be sold as the deeper thing.
Module 4: Machine learning, core, 6 to 8 weeks
- The supervised learning frame: features, targets, train and test splits, leakage
- Linear and logistic regression, k-nearest neighbours, decision trees, naive Bayes
- Evaluation: accuracy against precision and recall, F1, ROC AUC, and why accuracy misleads on imbalanced Indian datasets such as fraud or default
- Cross-validation and the bias-variance trade-off
- Unsupervised methods: k-means, hierarchical clustering, dimensionality reduction with PCA
Tooling is scikit-learn throughout. The test of this module is not how many algorithms it names but whether you can say why a model is wrong, which is what interviews probe.
Module 5: Machine learning, advanced, 4 to 6 weeks
- Ensembles: bagging, random forests, boosting, and gradient boosting with XGBoost or LightGBM
- Feature engineering and selection, encoding categorical variables, scaling
- Hyperparameter tuning with grid and randomised search
- Handling imbalanced classes, and choosing a metric that matches the business cost
- Time series basics, which recur constantly in Indian retail, lending and manufacturing work
SkilloVilla sells this ground as a standalone course, Machine Learning using Python at ₹38,110, which is worth knowing if machine learning is your only gap rather than the whole course.
Module 6: Deep learning and text analysis, 4 to 6 weeks
- Neural network fundamentals: layers, activation functions, loss, backpropagation
- Training in TensorFlow or PyTorch, and reading a training curve
- Convolutional networks for images at an introductory level
- Text: tokenisation, embeddings, sentiment and classification tasks
- Where large language models fit, and where a simpler model still wins
Most Indian job postings for a first data science role do not require deep learning. It belongs in the syllabus because it decides which second jobs are open to you, not because it gets you the first.
Module 7: Capstone and portfolio, 4 to 6 weeks
A capstone is only worth the weeks if it starts from a business question and a dirty dataset rather than a cleaned one with instructions attached. You should finish it able to explain, in two minutes and without notes, what you predicted, how well, and what the model gets wrong.
Three projects is the working portfolio minimum: one SQL-heavy analysis, one supervised model with a proper error analysis, one end-to-end piece where you chose the method yourself.
What a certificate syllabus leaves out, and a degree does not
Both routes teach machine learning. They differ in what sits underneath it and what you end up holding.
| Certificate or bootcamp | Degree | |
|---|---|---|
| Length | 6 to 12 months, as the providers above list | 3 to 4 years and up |
| Starts from | Python and business problems | Mathematics, statistics and programming foundations |
| Mathematics | Enough to use the methods | Taught as its own subject, and examined |
| Assessment | Projects, quizzes, a capstone | Proctored examinations, term by term |
| Cost shape | One fee, or instalments | Paid per term, with exit points |
| What you hold | A course certificate and a portfolio | A recognised academic qualification |
| Changes with the market | Quickly, which is why Generative AI modules appeared in 2026 | Slowly, by design |
The serious online degree route in India is the IIT Madras online BS in Data Science and Applications: the foundation level is ₹48,000, a BSc exit lands between ₹2,86,000 and ₹3,10,000, and the full BS between ₹3,86,000 and ₹4,50,000, paid per term across three to four years or more, largely self-paced with proctored exams, as listed in July 2026. The exit points are the useful part: you can stop at foundation level and still have something.
Pick the degree if you need the credential, for a research path, a government or PSU role, or further study. Pick the certificate if you need to be employable in under a year and already hold a bachelor’s degree in something. A certificate does not become a degree however good the teaching behind it is, and a degree does not make you job-ready on its own either.
What SkilloVilla’s data science track lists
The Data Science and AI track publishes its modules in order on its own page: Excel Beginner to Advanced, SQL Beginner to Advanced, Structured Problem Solving and Case Studies, Data Visualization in Power BI, Python Fundamentals, Statistics and Probability, Machine Learning Beginner to Intermediate, Machine Learning Advanced, Deep Learning and Text Analysis, and an AI workshop. It is listed at ₹1,01,999, currently ₹89,999, taught live with 1:1 mentorship, real projects and placement support. It runs about 8 months. Fees last checked September 2026; confirm current numbers with the provider before enrolling.
Notice the first four modules. They are an analytics syllabus, and they sit at the front deliberately, because a career switcher who cannot query a database and read a dashboard will not survive the machine learning block. The Data Science and Machine Learning track at ₹66,599, currently ₹53,299, drops Excel and Power BI and opens with Python fundamentals and analytics, then visualisation with matplotlib and seaborn, SQL, statistics and probability, and the same three machine learning modules. It suits someone who already codes.
For what these programmes cost across the market, our breakdown of data science course fees in India prices the tiers, and our shortlist of the best data science courses in India compares what each one actually teaches.
Five checks on any data science syllabus
- How many weeks go to statistics? Under three is an analytics course in disguise.
- Does machine learning include evaluation and error analysis, or only algorithm names?
- Is there a capstone with an unstructured problem, or guided exercises to the end?
- Are the tools current: scikit-learn, XGBoost, PyTorch or TensorFlow, rather than a list built in 2018?
- Does it assume Python you do not have? If the syllabus opens at pandas, check there is a foundations module before it.
Frequently asked questions
What subjects are in a data science course syllabus?
Python, SQL, statistics and probability, machine learning in two stages, deep learning and text analysis, and a capstone project. Career-switch programmes add Excel and a BI tool at the front so you can handle business data before you model it. The tools are pandas and NumPy, scikit-learn and XGBoost, and TensorFlow or PyTorch for the deep learning block.
How long is a data science course syllabus?
Published Indian programmes run six to twelve months: AlmaBetter lists 6 to 8 months, Intellipaat 7, upGrad’s IIIT Bangalore certificate 6 to 9, Coding Ninjas’ bootcamp 9, and Scaler and Great Learning 12, as listed in September 2026. A degree such as the IIT Madras online BS runs three to four years or more with exit points along the way.
Is a data science syllabus harder than a data analytics one?
It is longer and it demands more mathematics. The first four blocks, Excel, SQL, Python and visualisation, are shared. After that, analytics goes towards business framing and reporting while data science goes towards probability, statistical inference and model building, which is where people who skipped statistics get stuck. Neither is harder to get hired in; they are different jobs.
Do I need maths for a data science course?
Yes, more than an analyst needs. Probability, distributions and inference are used every week, and you need enough linear algebra and calculus to follow how a model is fitted rather than to derive it. School-level algebra plus a willingness to work through the statistics module is the realistic entry bar for a certificate programme.
Does a data science syllabus include Generative AI now?
Increasingly, yes, usually as a module at the end rather than as the spine of the course. Treat it as an addition to a syllabus that already covers statistics and machine learning properly, not as a replacement for them. A programme that leads with Generative AI and gives statistics one week is selling the headline.
Which is better, a data science certificate or a degree?
It depends on what the credential has to do. A certificate makes you employable in months and is judged on your portfolio, which suits anyone who already holds a bachelor’s degree. A degree is the right instrument if you need a formal qualification for research, further study, or roles that filter on it, and the IIT Madras online BS is the serious Indian route, with exit points if your plans change.
See the modules against a live track
If you want the syllabus above taught rather than assembled from videos, compare it with the SkilloVilla Data Science and AI track at ₹1,01,999, currently ₹89,999, which runs Excel, SQL, structured problem solving, Power BI, Python, statistics, two machine learning modules and deep learning with text analysis, taught live with 1:1 mentorship from working practitioners, real business projects and placement support, with scholarships based on your profile. Book a counselling call and ask for the module list and the cohort length in writing.
