Lecture 11: Statistical Estimation

So in pure statistics there’s just parameterized probability distributions and we have a parameter X and your job, you get one or more samples from one of these distributions and you’re charge is to say something intelligent about which distribution, which is to say which parameter value, generated the sample. So that’s statistics. So a standard technique is maximum likelihood estimations. In maximum likelihood estimation you do the following. You have an observation Y and you look at the density of the – you look at the density at Y or probability distribution if it’s a distribution on like – if its got different points on atomic points. ... See the whole transcript at [[http://see.stanford.edu/materials/lsocoee364a/transcripts/ConvexOptimizationI-Lecture11.pdf|Convex Optimization I - Lecture 11]]