3

I am working my way through linear models in matrix form using mainly Brown (2014) Linear Models in Matrix Form. The book presents a schematic representation of observed and fitted values in a linear regression model (a visualization I've often seen for linear regression) (copied from Brown 2014, p. 49):

Fig. 2.2. From Brown (2014) Linear Models in Matrix Form

While I understand the logic of this visualization, I wonder what the corresponding visualization is for generalized linear models when (i) the response variable is categorical (e.g. binary/ordinal), (ii) the response and predictors are categorical, (iii) there are multiple predictors. I have not seen such visualizations and find these as nice aids to grasp the intuition of regression models. Could anyone point me towards references for this?

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Wissenschaft
  • 599
  • 3
  • 15
  • 2
    Arthur Charpentier created some similar graphics for different models (e.g. Poisson) on [his blog](https://freakonometrics.hypotheses.org/9593). – COOLSerdash May 21 '20 at 11:15
  • 2
    example on logistic regression (binary outcome) can be found [here](https://stats.stackexchange.com/questions/259704/is-there-i-i-d-assumption-on-logistic-regression) – Haitao Du May 21 '20 at 11:29

0 Answers0