张俊妮《概率统计》北大光华-lecture26.pptVIP

Nominal Independent Variables in Multiple Linear Regression In many real-life situations one or more independent variables are nominal. Including nominal variables in a regression analysis model is done via dummy variables (哑变量). A dummy variable (I) can assume one out of two values, “zero” or “one”. Example: Car Price The dealer believes that color is a variable that affects a car’s price. Three color categories are considered: White Silver Other colors Note: Color is a nominal variable. Example: Car Price Point Prediction What is the point estimate of the mean selling price for all silver cars with odometer value of 40000? What is the point estimate of the selling price for an individual silver car with odometer value of 40000? Confidence Interval vs Prediction Interval What is the 95% interval estimate for the mean selling price for all silver cars with odometer value of 40000? What is the 95% interval estimate for the selling price for an individual silver car with odometer value of 40000? Confidence Interval vs Prediction Interval Question What does it mean for y’=log(y) to decrease by 0.00022 unit? More subtle case: one of the independent variables can almost be expressed as linear combinations of other independent variables. We can regress each independent variable on the other independent variables, and if the resulting R-square is high, then there exists multicollinearity. Theoretical Reason: If there is a multicollinearity problem, will be close to a singular matrix, therefore the variances of the estimators for the coefficients will be large Multicollinearity causes two kinds of difficulties: The t statistics appear to be too small. The coefficients cannot be interpreted as “slopes”. More on the Multicollinearity Issue dfghdfghfdghgfhdfghdfgh * dfghdfghfdghgfhdfghdfgh * * * Lecture 26 Qualitative Independent Variables in Linear Regression Prediction Analysis of Residual The Multicollinearity Issue * Classification o