Description
To help users migrate from Spark 1.6. to 2.0, we should make model loading backward compatible with models saved in 1.6. The main incompatibility is the vector column type change.
Attachments
Issue Links
- contains
-
SPARK-16133 model loading backward compatibility for ml.feature
- Resolved
-
SPARK-16240 model loading backward compatibility for ml.clustering.LDA
- Resolved
-
SPARK-16130 model loading backward compatibility for ml.classfication.LogisticRegression
- Resolved
-
SPARK-16132 model loading backward compatibility for tree model (DecisionTree, RF, GBT)
- Resolved
-
SPARK-16177 model loading backward compatibility for ml.regression
- Resolved
-
SPARK-16241 model loading backward compatibility for ml NaiveBayes
- Resolved
-
SPARK-16243 model loading backward compatibility for ml.feature.PCA
- Resolved
-
SPARK-16245 model loading backward compatibility for ml.feature.PCA
- Resolved
-
SPARK-16176 model loading backward compatibility for ml.recommendation
- Closed
- relates to
-
SPARK-15573 Backwards-compatible persistence for spark.ml
- Resolved