Inferring properties of an underlying probability distribution (a data-generating process) by means of random samples.
Can be done via frequentist methods (frequentist statistics) or Bayesian methods (Bayesian statistics). Can be done for a fixed-number of samples (fixed-time) or sequentially (safe, anytime-valid inference (SAVI)).
Does it feel stupid to have a page dedicated to such a fundamental concept? Yes, yes it does. But hey, Cosma Shalizi has a note on statistics full stop, and he’s much more accomplished than me. Hopefully this note will also be full of hot takes someday.