Regression diagnostics

مباحث تشخيصي رگرسيون

Tests to identify the main problem areas in regression analysis: normality, common variance and independence of the error terms; outliers, influential data points, collinearity, independent variables being subject to error, and inadequate specification of the functional form of the model. The purpose of the diagnostic techniques is to identify weaknesses in the regression model or the data. Remedial measures, correction of errors in the data, elimination of true outliers, collection of better data, or improvement of the model, will allow greater confidence in the final product. See also error terms, residuals (including likelihood distance test), leverages and Cook statistics.