6b7527e381
### What changes were proposed in this pull request? In https://github.com/apache/spark/pull/21632/files#diff-0fdae8a6782091746ed20ea43f77b639f9c6a5f072dd2f600fcf9a7b37db4f47, a new field `rawCount` was added into `NodeData`, which cause that a tree model trained in 2.4 can not be loaded in 3.0/3.1/master; field `rawCount` is only used in training, and not used in `transform`/`predict`/`featureImportance`. So I just set it to -1L. ### Why are the changes needed? to support load old tree model in 3.0/3.1/master ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? added testsuites Closes #30889 from zhengruifeng/fix_tree_load. Authored-by: Ruifeng Zheng <ruifengz@foxmail.com> Signed-off-by: Sean Owen <srowen@gmail.com> |
||
---|---|---|
.. | ||
benchmarks | ||
src | ||
pom.xml |