glm will not converge

[jblakesl]

What kinds of data cannot be put into a glm? I have tried everything to get my regression model to converge, but it will not. I have a dataset with 82 variables. Does the data have to be continuous? I cleaned the data and tried maxit=100.

[mattwarkentin]

Can you share the data you are trying to use or provide a reproducible example? Can you at least provide the error message you are receiving?

[jblakesl]

Maybe you can help me with my code too! I am trying to predict whether a customer liked the meal or not in a restaurant. So it is binary 1=satisfied , 0=unsatisfied.

DATA:

name_id	order_id	   food_type	satisfied_unsatisfied	
582	      115688  	        meat	                  0	
582	     115689	        meat	                  1	
582	     115691	        meat	                  0	
582	     115692	        meat	                  0	
8	     1384	                salad	                  0
8	      1385	                meat	                  0
8	      1386	                meat	                  0
262	      1387	                meat	                  0
262	      1388	                meat	                  0
262	      1389	                salad	                  0

Here is the code I am running and am not getting a percentage between 0 and 1 for predictions of satisfaction. What should be in the column for your test dataset for the predictions to replace? Is it blank or NA?

CODE:

library(ggplot2)
library(dplyr)
train=read.csv("C:\\Users\\jb\\Downloads\\Food_Training2.csv")
attach(train)
head(train)
###Predict Satisfaction####
train.clean=na.exclude(train)
model=glm(satisfied_unsatisfied~.,data = train.clean,family = binomial(link = "logit")) 
summary(model)
p.sat=predict(model,newdata=train.clean,type="response")
test=read.csv("C:\\Users\\jb\\Downloads\\Food_Testing2.csv")
p1=predict(model,test)
p.off=rep(NA,nrow(test))
data=data.frame(order_id=1:nrow(test),satisfied_unsatisfied=p.sat)