Step 2
Probability and Statistics
The language of working with uncertainty: probability, distributions, mean and spread, hypothesis tests. Every job that touches data and every machine learning model depends on it.
Know these firstDiscrete Mathematics
- Probability rules
- Conditional probability and Bayes
- Random variables
- Normal and other distributions
- Expected value and variance
- Confidence intervals
- Hypothesis testing
Why it matters
Is an A/B test result real or luck, how large is a model's error, and what does a classifier mean when it says "90% sure"? This course answers all of these. It is the maths that data analysts, data scientists and AI engineers use most.
What to know
- Probability rules, counting and independence.
- Conditional probability and Bayes' rule; the false positive example in medical tests.
- Discrete and continuous random variables; binomial, Poisson and normal distributions.
- Expected value, variance and standard deviation.
- The law of large numbers and the central limit theorem.
- Estimating from a sample: standard error and confidence intervals.
- Hypothesis testing, p-values and what they do not mean.
- Correlation and simple linear regression.
Practice and projects
- Find a probability you do not know the formula for by simulating it thousands of times, then calculate it and compare.
- Take an open dataset (e.g. from Kaggle), compute means, spreads and distribution plots, and compare two groups with a hypothesis test.
Common mistakes
- Mistaking correlation for causation.
- Reading a p-value as "the probability that the hypothesis is true".
- Looking only at the mean and missing the distribution and outliers.