Saltar al contenido ForoHackers

Probability and statistics: start here

Probabilidad y estadística
#633 · Ophuscado · · Denunciar

Probability describes uncertainty with numbers; statistics goes the other way, using data to learn about the process that produced it. Computing relies on both: randomised algorithms, hash tables, A/B tests, spam filters, performance measurements and every machine-learning model. The skill is to reason carefully about chance instead of trusting intuition.

Key concepts

  • Random variables and distributions: a quantity with uncertain value, described by its distribution, such as binomial, normal or exponential.
  • Expectation and variance: the average value and the spread around it.
  • Conditional probability and Bayes' rule: how evidence updates belief.
  • Law of large numbers and central limit theorem: why averages settle down and look normal.
  • Inference: estimation, confidence intervals and hypothesis tests, and the Bayesian alternative.

Vocabulary

  • Independence: knowing one event tells nothing about the other.
  • Sample and population: the data you have, and the whole you want to describe.
  • p-value: how surprising the data would be if there were no real effect; it is not the probability that the hypothesis is true.
  • Bias and variance: systematic error, and sensitivity to the particular sample.

What to study, in order

  1. Counting, events, conditional probability and Bayes' rule, with a medical-test example.
  2. Simulate dice, coins and random walks in code and compare results with formulas.
  3. Study the common distributions and verify the central limit theorem by simulation.
  4. Estimate a quantity with a confidence interval, then run a hypothesis test and say what it does not show.
  5. Learn regression, then see how it connects to learning from data.

Related tags

#combinatorics #data-science #machine-learning #simulation #information-theory

Inicia sesión o únete para responder.