Formula Cheat Sheet
Everything in one place.
Here are the most important formulas from the whole course, in one place. Pause on each table, and maybe take a screenshot to keep.

- Every key formula
- Pause on each table
- Keep it handy
Describing and probability
The mean is the sum divided by N. The variance is the average squared distance from the mean, and the standard deviation is its square root. A z score is the distance from the mean in standard deviations. Bayes' theorem flips a conditional probability. And the standard error is the standard deviation over the square root of N.
| Idea | Formula |
|---|---|
| mean | Σx ÷ n |
| variance | Σ(x − x̄)² ÷ (n − 1) |
| z-score | (x − mean) ÷ std |
| Bayes | P(A|B) = P(B|A)·P(A) ÷ P(B) |
| standard error | std ÷ √n |
| 95% CI | mean ± 1.96 × SE |
Math for machine learning
The dot product multiplies and sums. Cosine similarity divides it by both lengths. Gradient descent steps against the gradient. The sigmoid squashes into a probability. Cross entropy is minus the log of the correct probability. And ridge regression adds the sum of squared weights to the loss.
| Idea | Formula |
|---|---|
| dot product | Σ aᵢbᵢ |
| cosine | a·b ÷ (‖a‖‖b‖) |
| gradient descent | w ← w − lr × ∇L |
| sigmoid | 1 ÷ (1 + e⁻ᶻ) |
| cross-entropy | −log p(correct) |
| ridge | loss + α Σ w² |
Summary
That's the whole course on two tables. Describing data, probability, inference, and the math inside models. Next, the final quiz.
Key takeaways- Describe and infer
- Vectors and gradients
- Losses and penalties