Mon premier blog

Regression Analysis of Count Data A. Colin Cameron
Publisher: Cambridge University Press

Bar some exceptions, most big data insights today are based on simple counting, linear correlations or at best based on impoverished models like linear regression. Regression analysis - in it's generality is powerful. Since the outcome variable “absenteeism” is a count variable, Poisson, Quasi-Poisson, Negative binomial and Zero inflated models are applied and compared on the basis of Log likelihood, AIC, regression coefficients and standard errors of the best fit. One competitive and one noncompetitive. Bivariate analysis and logical regression models were unsatisfactory. This recent article [2] in BJD explores the concept of Polysensitisation (PS) in contact dermatitis They have used a negative binomial hurdle regression method for count data to independently estimate risk to be sensitised at all and the risk of having several contact allergies, i.e., to be polysensitised. The course also covers new classes of models for binary and count data, emphasizing the need to fit appropriate models to the underlying processes generating the data being explained. Pertinent refs: http://cameron.econ.ucdavis.edu/racd/count.html and the book by the same authors, A.C.Cameron, P.K.Trivedi, REGRESSION ANALYSIS OF COUNT DATA (1998). It was found For example, in social data analysis, Poisson regression models were used to assess the effects of parental and peer approval of smoking on adolescents' current level of smoking (Siddiqui et al., 1999). We used paired data analysis to compare discrepancies between poll and official count for these matched pairs. It should also be noted that a regression analysis of magnitude/direction of shift relative to magnitude of contest margin yields an F value of 21.9, corresponding to a p value of p<0.000022 and strongly corroborating our finding of strong correlation using the paired testing approach.