Note: reconstruct_forest (reconstructForest) deserialization logic is changed as well as Python part of serialization: tree_counts contains 3 values for each tree instead of 4: numNodes, numSplitIdx, numAvIdx
Verified by adding test_pickle_unpickle.py, also sklearn test set contains its own pickle - unpickle test
Note: reconstruct_forest (reconstructForest) deserialization logic is changed as well as Python part of serialization: tree_counts contains 3 values for each tree instead of 4: numNodes, numSplitIdx, numAvIdx Verified by adding test_pickle_unpickle.py, also sklearn test set contains its own pickle - unpickle test
Issue https://github.com/forestry-labs/Rforestry/issues/98
Continuation of https://github.com/forestry-labs/Rforestry/pull/101 (closed in favour of this one)