University of Colorado Boulder ASEN 5070: Statistical Orbit Determination I Fall 2015 Professor Brandon A. Jones Lecture 11: Probability and Statistics (Part 1)
University of Colorado Boulder Lecture Quiz 4 Due 5pm ◦ Due by 5pm on Friday Homework 4 Due September 25 ◦ As mentioned in , it is different from one originally posted at the start of the semester Final Make-up Lecture 4pm 2
University of Colorado Boulder Axioms of Probability Probability Distributions Multivariate Distributions 3
University of Colorado Boulder 4 Axioms of Probability
University of Colorado Boulder X is a random variable (RV) with a prescribed domain. x is a realization of that variable. Example: ◦ 0 < X < 1 ◦ x 1 = ◦ x 2 = ◦ x 3 = ◦ etc 5
University of Colorado Boulder The conceptual definition holds for a discrete distribution Requires more mathematical rigor for a continuous distribution (more later) 6
University of Colorado Boulder Probability of some event A occurring: Probability of events A and B occurring: Axioms: 7
University of Colorado Boulder 8
University of Colorado Boulder Although we often see a probability written as a percentage, a true mathematical probability is a likelihood ratio 9
University of Colorado Boulder Mathematical definition of conditional prob.: 10 Example:
University of Colorado Boulder Two events are independent iff 11 Why is the latter true if A and B are independent?
University of Colorado Boulder 12 Probability Distributions
University of Colorado Boulder Random variables are either: ◦ Discrete (exact values in a specified list) ◦ Continuous (any value in interval or intervals) Examples of each: ◦ Discrete: ◦ Continuous: 13
University of Colorado Boulder DRVs provide an easier entry to probability 14 They are vary important to many aerospace processes! However, StatOD tends to deal more with CRVs ◦ Rarely discretize the system of coordinates We will primarily discuss the latter!
University of Colorado Boulder Probability of X in [x,x+dx]: 15 where f(x) is the probability density function (PDF) For CRVs, the probability axioms become:
University of Colorado Boulder 16 Using axiom 2 as a guide, how would we derive k in the following:
University of Colorado Boulder For the cases X ≤ x, let F(x) be the cumulative distribution function (CDF) 17 It then follows that: ?? ?
University of Colorado Boulder 18 From the definition of the density and distribution functions we have: From axioms 1 and 2, we find:
University of Colorado Boulder 19 Multivariate Distributions
University of Colorado Boulder The PDF for two RVs may be written as: 20 Hence, for two RVs:
University of Colorado Boulder How do we compute probabilities given a multivariate PDF? 21
University of Colorado Boulder We often want to examine probability behavior of one variable when given a multivariate distribution, i.e., 22 Marginal density fcn of X
University of Colorado Boulder What would be the marginal probability density function of Y? 23
University of Colorado Boulder What if I only care about the probability of one variable? 24 Alternatively,
University of Colorado Boulder Analogous to definition previously discussed, but rooted in the PDFs and marginal distributions 25
University of Colorado Boulder If X and Y are independent, then 26 ?? ?