XGBoost in final model what does "yes=3,no=4" mean?

Question

XGBoost v0.6 Binary Classification

$ cat predin.txt 
0 1:10 2:10 3:10
1 1:10 2:10 3:100
1 1:10 2:100 3:10
1 1:100 2:10 3:10

In other words, the negative instance label 0 requires low values for all three features.

./xgboost/xgboost m.conf
./xgboost/xgboost m.conf task=dump model_in=0002.model name_dump=dump.raw.txt
$ cat dump.raw.txt 
booster[0]:
0:[f1<55] yes=1,no=2,missing=1
    1:[f2<55] yes=3,no=4,missing=3
        3:[f3<55] yes=5,no=6,missing=5
            5:leaf=-1.5
            6:leaf=1.5
        4:leaf=1.5
    2:leaf=1.5
booster[1]:
0:[f1<55] yes=1,no=2,missing=1
    1:[f2<55] yes=3,no=4,missing=3
        3:[f3<55] yes=5,no=6,missing=5
            5:leaf=-0.784694
            6:leaf=0.784694
        4:leaf=0.784694
    2:leaf=0.784694

What does "yes=3,no=4" mean?

Answer 1

I assume it should read as follows:

• if condition is true, go to branch 3
• otherwise go to branch 4

I am only guessing, but seems plausible