Probability you'll actually use
Take this if probability at school felt like a topic about coins and dice rather than something anyone uses.
What you’ll be able to do
You can update a belief correctly when new evidence arrives, and explain why a 99% accurate test for a rare disease still mostly returns false alarms.
What it assumes
You are comfortable with fractions and percentages. School probability helps but is not assumed.
- Price
- Free
- Length
- 1 week
- Sessions
- 4
- Effort
- ~2 h/week · 2 h total
- For
- Anyone
Adds it to My courses on your dashboard. No card, nothing to cancel.
What’s taught, in order
- 01
A belief with a number on it
Writing down how strongly you believe something before seeing evidence, which is the prior. Committing to a number first is what makes the update in the next module feel like something rather than algebra.
- 02
Updating on evidence
Working through the rare-disease test by hand, getting the answer wrong the way almost everyone does, and then seeing where the intuition failed. The mistake is the lesson.
- 03
The shape most things have
Why so many measurements pile up in the middle, what the standard deviation actually marks off, and how to read an unusual event off the shape rather than guessing.
- 04
A spam filter, by hand
Combining priors and evidence across several words to classify one message. This is naive Bayes in full, done on paper, with no library and nothing hidden.
What you build
Take ten real messages from your own inbox, five you would call spam and five you would not. Pick three words that discriminate between them, compute the priors by hand, then classify a fresh message and check whether you agree with your own filter.
Check it yourself against this
- Your three words are ones that actually differ in frequency between the two piles.
- You computed the update by hand at least once without a calculator doing the reasoning for you.
- Your filter's verdict on the fresh message is one you can justify from the numbers.
- You can explain why adding a fourth word could make the filter worse rather than better.
Then answer this
The disease test is 99% accurate and your result is positive, yet you probably do not have the disease — what is the 99% actually measuring?
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 the working half of probability, not the subject: no random variables in general, no distributions beyond the normal one, no proofs, and no continuous maths. It gives you enough to reason about evidence and to follow a naive Bayes classifier, and stops there.
Where this leads
The Bayes visualiser is free and standalone, and it is the quickest way to see the rare-disease result for yourself — drag the base rate down and watch the false alarms swamp the true positives.
FAQ
- Is this useful for board exams?
- It overlaps the probability chapters in Class 11 and 12 but does not follow their sequence or their question style, so treat it as understanding rather than revision. The Class 12 Maths to Machine Learning sprint is the one built around the board syllabus.
- Do I need this before other AI courses?
- Not required, but it makes several later ideas much easier — evaluation metrics, model confidence, and anything involving uncertainty all assume you can reason about a prior. It is a good second free course after the gradient descent one.
- Why by hand rather than in code?
- Because a library call hides the exact step where intuition breaks. Once you have got the rare-disease answer wrong on paper and seen why, you will never mis-read a precision or recall number again, and that is worth more than a working script.
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.