quadratic discriminant analysis in R "some group is too small for qda"

Question

when I run qda using a certain number of variables but when I try to add a new one it generates the error i tried checking the variables and they satisfy the conditions when I remove one of the already used variables with the new one that I wanted to add it works and when I plug the old one back in it generates the same error.

    model=qda(data=males,ZONE~log(L)+log(LT)+log(HG)+log(HD)+log(HS)+log(LB)+log(LH)+log(LP)+log(PP)+log(PF))
#this code works fine

#adding a new variable from the same dataset and it's numeric 
model_2=qda(data=males,ZONE~log(L)+log(LT)+log(HG)+log(HD)+log(HS)+log(LB)+log(LH)+log(LP)+log(PP)+log(PF)+log(TP))

#this generates the error 'some variable is too small for qda
#now i replace one of the existing variables with the one i wanted to add
model_3=qda(data=males,ZONE~log(L)+log(LT)+log(HG)+log(HD)+log(HS)+log(LB)+log(LH)+log(LP)+log(PP)+log(TP))
## i replaces log(PF) with log(TP) and it worked so the problem is not in the variable

any suggestions of what might be causing this?

how many variables and how many samples do you have ? can you post your data? or a portion of your data here ? — user1267127
i don't know the appropriate way to post data as this is my first post on the site. i have 63 rows and 35 columns 14 of which are categorical and 19 numerical (continuous) and i didn't use any categorical variables in the model. so far i've been able to use only 10 variables — oussama_ag

Kristoffer Vitting-Seerup Kristoffer Vitting-Seerup · Accepted Answer · 2018-04-04T12:02:11

The problem seems to be that you have more variables (columns) than the smallest of your classes. You need to reduce the number of variables or use a LDA instead.

quadratic discriminant analysis in R "some group is too small for qda"

1 Answers