Random variables are formalised on measure-theoretic grounds before discrete and continuous distributions, expectation, variance and covariance are simulated with numpy.random. Frequentist inference (MLE, hypothesis testing, confidence intervals) is paralleled with Bayesian updating, and resampling techniques such as bootstrap and permutation tests provide non-parametric robustness. The chapter closes with time-series basics and goodness-of-fit diagnostics, preparing the reader for stochastic modelling in later chapters.