Class 12 maths to machine learning
Take this if you are revising Class 12 calculus anyway and would rather learn it once, properly, with a reason attached.
What you’ll be able to do
You can differentiate a real loss function by hand, apply the chain rule through two composed steps, and write the code that uses the result to improve a model.
What it assumes
You are studying or have studied Class 11 calculus — limits and basic derivatives. No coding experience assumed; Python is introduced as needed.
- Price
- ₹299
- Length
- 4 weeks
- Sessions
- 12
- Effort
- ~4 h/week · 16 h total
- For
- Class 11–12
Adds it to My courses on your dashboard so you can find it again. Payment is not open yet — nothing is charged.
What’s taught, in order
- 01
Derivatives as sensitivity
The board definition is a limit; the useful definition is how much the output moves when I nudge the input. Both are the same thing, and holding them together is what makes the rest of the sprint work.
- 02
The chain rule, as a pipeline
Composed functions taught as a chain of nudges passing backwards through a pipeline. This is the exact structure of backpropagation, met four years before most people meet it.
- 03
Partial derivatives and the gradient
Two knobs instead of one, and the vector of both. The board treats this as an extension of a technique; here it is the object that every training loop actually computes.
- 04
Closing the loop in code
Writing the twenty lines that take your hand-derived gradient and use it to fit a line to real data. Watching your own derivative reduce a real error is the payoff the syllabus never gives you.
What you build
Derive the gradient of mean squared error for a straight-line model by hand, on paper, showing the chain rule step explicitly. Then implement exactly that formula in Python and fit it to a dataset you choose, and confirm your hand derivation and your code agree to three decimal places.
Check it yourself against this
- Your derivation shows the chain rule step, not just the final expression.
- Your code computes the gradient from your own formula, not from a library's autodiff.
- Your fitted line reduces the error from its starting value, and you can show the curve.
- Hand-computed and code-computed gradients agree at a test point to three decimal places.
- You can state what each partial derivative is telling you about its own parameter.
Then answer this
Your board textbook and this sprint teach the same chain rule, so why does it feel like a different topic here?
Two or three sentences, in your own words. If you cannot, the course is not finished — go back to the module it came from.
What this course does not do
This is not board exam coaching and does not follow the CBSE or ISC question format, so it will not by itself prepare you for the paper. It also covers only the calculus that machine learning needs — integration by parts, definite integral applications and differential equations are not here.
Where this leads
This sprint is a four-week slice of the twenty-week Foundations phase, which takes the same approach across all of calculus, linear algebra, probability and Python rather than just the calculus you need for a first gradient.
FAQ
- Will this help my board marks?
- Indirectly. Students who understand what a derivative is for tend to make fewer mechanical errors, but this sprint teaches understanding rather than exam technique and never drills past-paper questions. Treat it as the thing that makes revision make sense, not as revision.
- I am in Class 11. Should I wait?
- You can start once you have met derivatives, which is usually mid-Class 11. The chain rule module assumes you have seen composed functions; if you have not, take the free gradient descent course first and come back.
- Do I need to install anything?
- No. The Python runs in your browser with nothing to set up, which matters when you are working on a school machine or a shared laptop. If you would rather use your own local Python, the code is plain and portable.
Other courses
This is a standalone course. It does not add weeks to the 78-week curriculum or change your roadmap. If you want the whole path instead, the full programme is ₹399/month · ₹2,000/year — see what it covers.