Slide 1OverviewOverviewPROBABILITY AND STATISTICS IN COMPUTER SCIENCE AND SOFTWARE ENGINEERING Chapter 10: Statistical Inference – part 21OVERVIEWLast time we saw that we could expand our hypothesis testing idea to test if a population had a particular distributionThis is called “goodness of fit” testingUsually, we do not know the parameters of the underlying population, but we can use our sample data estimate those parametersThis process involved dividing the range of possible values for the random variable into “bins” (preferably 5-8), and then comparing the expected number of observations (according to the Null hypothesis) to the observed number in each binWe used a -squared test, and these were right-tailed tests•<2OVERVIEWWe also saw how to use similar ideas to test independence between factors in a populationAgain, this involved a testIn this lecture, we will discover tests that will allow us to conduct non-parametric tests for things like a population medianThese tests will involve ranking the data and computing P-values much like we did before, but with new distributionsWe’ll introduce the signed test and Wilcoxon signed rank test to accomplish
View Full Document