GLM diagnostic plots in r

#1
I am trying to fit a Gamma glm in R. getting very weird diagnostics even though all variables are significant. Please advise as to how i can improve this
 
#2
I am trying to fit a Gamma glm in R. getting very weird diagnostics even though all variables are significant. Please advise as to how i can improve this
It would appear that there is a distinct group in your data, that you are not fitting. This can happen when you don't include all variables. Can you identify whether that cluster below conforms to any group?
 
#3
It would appear that there is a distinct group in your data, that you are not fitting. This can happen when you don't include all variables. Can you identify whether that cluster below conforms to any group?
Thanks for your reply, it helps. I understand that there might be regional/geographical differences in the data. but how do i find that group ?
 
#4
Thanks for your reply, it helps. I understand that there might be regional/geographical differences in the data. but how do i find that group ?
Add the residuals to your dataset (as a column), an then subset the data to residuals < -6. That should effectively isolate that group in the data, then hopefully you can find out what's up.

You can extract the residuals with residuals(mymodel)