I appreciate your help.
In Random forest first:
- define number of trees
- partition data by bootstrapping
- on each partition construct trees (in each node a sub sample of features is selected)
- label leaf nodes
- for classifying a new instance vote over all trees.
- Partition dataset
- Build AVC-set of a partition
- Build tree over the partition by computing a purity criterion (like gini-index) over AVC-sets