Regression-Analysis-on-Drinking-Data

• “Cirrhosis death rate” as the response and “Urban population”, “Late births”, “Wine consumption per capita”, “Liquor consumption per capita” as the covariates.
• Fitting OLS, Least Absolute Deviation, Least Median Square, Least Trimmed Squares Estimate. Selecting model by R 2 , AIC, BIC, LOOCV. Homoskedasticity check residuals vs fitted plot, BP test; Correlation ACF, PACF plot, Durbin-Watson Test; Detecting Influential Points Studentized Residuals, DFBETAS, DFFITS, COVRATIO, Cook’s D.
• Improvement in the normality assumptions of the residuals after removing Influential Points. Normality check QQ plot, Shapiro-Wilk Test. Multicollinearity check by VIF. “Urban population”, “Liquor consumption per capita” explain most of the linearity in “Cirrhosis death rate”.