Understanding Distribution Parameters and Expectation in Statistics
130 likes | 281 Vues
Learn about expectation, moments, variance, and other distribution parameters in statistical analysis. Explore MGF, characteristics of distributions, and functions. Understand mixed distributions and their key properties.
Understanding Distribution Parameters and Expectation in Statistics
E N D
Presentation Transcript
Expected Value (mean) • As the number of trials increases, the average outcome will tend towards E(X): the mean • Expectation: • Discrete • Continuous
Expectation of h(x) • Discrete • Continuous
Moments of a Random Variable • n: positive integer • n-th moment of X: • So h(x) = X^n • Use E[h(X)] formula in previous slide • n-th central moment of X (about the mean): • Not as important to know
Variance of X • Notation • Definition:
Important Terminology • Standard Deviation of X: • Coefficient of variation: • Trap: “Coefficient of variation” uses standard deviation not variance.
Moment Generating Function (MGF) • Moment generating function of a random variable X: • Discrete: • Continuous:
Two Ways to Find Moments • E[e^(tx)] 2. Derivatives of the MGF
Characteristics of a Distribution • Percentile: value of X, c, such that p% falls to the left of c • Median: p = .5, the 50th percentile of the distribution (set CDF integral =.5) • What if (in a discrete distribution) the median is between two numbers? Then technically any number between the two. We typically just take the average of the two though • Mode: most common value of x • PMF p(x) or PDF f(x) is maximized at the mode • Skewness: positive is skewed right / negative is skewed left • I’ve never seen the interpretation on test questions, but the formula might be covered to test central moments and variance at the same time
Expectation & Variance of Functions • Expectation: constant terms out, coefficients out • Variance: constant terms gone, coefficients out as squares
Mixture of Distributions • Collection of RV’s X1, X2, …, Xk • With probability functions f1(x), f2(x), …, fk(x) • These functions have weights (alpha) that sum to 1 • In a “mixture of distribution” these distributions are mixed together by their weights • It’s a weighted average of the other distributions
Parameters of Mixtures of Distributions • Trap: The Variance is NOT a weighted average of the variances • You need to find E[X^2]-(E[X)])^2 by finding each term for the mixture separately