0
$\begingroup$

The response variable is FreeKick (the duaration for completing a freekick in football match,unit is second,double), and explanatory variables is Timing (the timing for the freekick tooking place,Continuous variable,unit is second,double), xZone(the zone for the freekick tooking place, factor,6 level),gameState(the goal difference for one team in football match, 5 level). First, we conduct a linear model.

MFK1<-lm(FreeKick ~ Timing*fxZone*gameState, data=freekick)

enter image description here Obviously, this model is not good, cause its residuals violates the normality assumptions and with too many outliers. My question is can i transfer to quasi-poisson GLM? For the histogram of FreeKick is similary with some kind of poisson distribution. I am not sure that if poisson base GLM fits for the non-count data(FreeKick).Can i anlysis the interaction effect in GLM as below? I am newer to it.Thanks for your suggestion.

MFKglm<- glm(FreeKick ~ Timing*fxZone*gameState,family=quasipoisson, data=freekick)

enter image description here

$\endgroup$
  • $\begingroup$ I think that the exponential distribution might be a good option in your case. See there: stats.stackexchange.com/a/240459/11849 $\endgroup$ – Roland Jun 4 at 7:52
  • $\begingroup$ appreciate for your reply. I just wonder how to report the interaction effect in Gamma GLM.Thanks in advance. $\endgroup$ – Yang-Qing Zhao Jun 5 at 1:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.