GBM error in classification bernoulli distribution

Question

When running the gbm function for a classification problem. I get the following error:

Error in res[flag, ] <- predictions : replacement has length zero

I would like to know why I get this error and how to solve it.

My data is about 77 numeric variables(intergers) to be used in the classification and the 1 grouping factor. No other variables are in the data. There is no missing data in the data. The grouping factor is coded as a factor (0,1) as required.

The structure of my data looks something like this:

$Group : Factor w/ 2 levels "0", "1"
$it1 : int
...
$it70 : int

my model looks like this:

mod_gbm <- gbm(Group~. distribution = "bernoulli", data=df,
               n.trees=1000,shrinkage=.01, n.minobsinnode=5, 
               interaction.depth = 6, cv.folds=5)

I realize this question is very similar to the one here: Problems in using GBM function to do classification in R but that person was wondering about using a numeric variable and the only response was to remove cv.folds. I would like to keep cv.folds in my model and to have it run.

Going to need to try and figure out if the issue is with your dataset or the gbm package. Does classification work for a single decision tree? Could you try a different package which has a boosting implementation? Is there any error when the model is trained? — Jamie_B, Apr 30 '20 at 11:19
I cannot run the model with 1 tree but I can run the gbm3 package with no error. — Heather Clark, May 03 '20 at 13:08

score 1 · Accepted Answer · answered Jun 25 '20 at 20:54

If you check out the vignette of gbm:

distribution: Either a character string specifying the name of the
          distribution to use or a list with a component ‘name’
          specifying the distribution and any additional parameters
          needed. If not specified, ‘gbm’ will try to guess: if the
          response has only 2 unique values, bernoulli is assumed;
          otherwise, if the response is a factor, multinomial is
          assumed

If you only have two classes, you don't need to convert it into a factor. We can explore this with iris example, where I create a group label 0/1 :

library(gbm)
df = iris
df$Group = factor(as.numeric(df$Species=="versicolor"))
df$Species = NULL
 
mod_gbm <- gbm(Group~.,distribution ="bernoulli", data=df,cv.folds=5)
Error in res[flag, ] <- predictions : replacement has length zero

I get the same error. So we convert it to numeric 0/1 and you can see it works correctly.

When the variable is a factor, doing as.numeric() converts it to 1,2 with 1 corresponding to the first level. So this case, since Group is 0/1 to start with:

df$Group = as.numeric(df$Group)-1
mod_gbm <- gbm(Group~.,distribution ="bernoulli", data=df,cv.folds=5)

And we get the predictions:

pred = ifelse(predict(mod_gbm,type="response")>0.5,1,0)
table(pred,df$Group)

    
pred  0  1
   0 98  3
   1  2 47

GBM error in classification bernoulli distribution

1 Answers1

Linked