MGBU 2142 Lecture Notes - Lecture 16: Prediction Interval, Confidence Interval, Bias Of An Estimator
4 Nov 2016
School
Department
Course
Professor
Document Summary
Assumptions of simple linear regression model: in terms of dependent variable. Standard deviations of distribution of y values around their mean for any. Describes our data, telling us magnitude of our relationship. Permits us to make forecast of y for given value of x: equation is (cid:1877)=(cid:2868)+(cid:2869)(cid:1876)+ mean being (cid:2868)+(cid:2869)(cid:1876) All values of normally distributed w/=0 and standard deviation of. The standard deviations of are the same regardless of the value of x. Y values statistically independent of one another: in terms of errors (homoscedasticity) Can we improve over minimum absolute errors: can improve over minimum absolute errors by using minimum squared errors, which gives best linear unbiased estimators & maximum likelihood estimator. How do we find line that minimizes the sum of squares: technically, take partial derivatives of sum of squared errors with respect to b0 and b1. Confidence interval for true mean of y.
