[SPARK-4789] [SPARK-4942] [SPARK-5031] [mllib] Standardize ML Prediction APIs#3637

Closed

jkbradley wants to merge 30 commits intoapache:masterfrom jkbradley:ml-api-part1

+1,753-156

Commits on Feb 5, 2015

Added lots of classes for new ML API:
jkbradley
committed
fixed compilation issues, but have not added tests yet
jkbradley
committed
removing everything except for simple class hierarchy for classification
jkbradley
committed
Added LinearRegression and Regressor back from ml-api branch
jkbradley
committed
Modified ParamMap to sort parameters in toString. Cleaned up classes in class hierarchy, before implementing tests and examples.
jkbradley
committed
Fixed bug from last commit (sorting paramMap by parameter names in toString). Fixed bug in persisting logreg data. Added threshold_internal to logreg for faster test-time prediction (avoiding map l…
jkbradley
committed
Fixed issue with logreg threshold being set correctly
jkbradley
committed
Updated docs. Added LabeledPointSuite to spark.ml
jkbradley
committed
* Changed semantics of Predictor.train() to merge the given paramMap with the embedded paramMap.
jkbradley
committed
added train() to Predictor subclasses which does not take a ParamMap.
jkbradley
committed
* fixed LinearRegression train() to use embedded paramMap
jkbradley
committed
Added JavaLabeledPointSuite.java for spark.ml, and added constructor to LabeledPoint which defaults weight to 1.0
jkbradley
committed
Added methods:
jkbradley
committed
Added spark.ml LinearRegressionSuite
jkbradley
committed
rat fix
jkbradley
committed
* Made ProbabilisticClassificationModel into a subclass of ClassificationModel. Also introduced ProbabilisticClassifier.
jkbradley
committed
Fixed bugs from previous commit.
jkbradley
committed
small cleanup
jkbradley
committed
Fixed Linear/Logistic RegressionSuites
jkbradley
committed
Fixed bug in LogisticRegression (introduced in this PR). Fixed Java suites
jkbradley
committed
added blanket mima exclude for ml package
jkbradley
committed
Updates based on code review. Major ones are:
jkbradley
committed
fixed after sql datatypes PR got merged
jkbradley
committed
Added MIMA excludes for VectorUDT (now public), and added DeveloperApi annotation to it
jkbradley
committed
fixed JavaLinearRegressionSuite.java Java sql api
jkbradley
committed
Fixed issues after rebasing from master (after move from SchemaRDD to DataFrame)
jkbradley
committed
fixed test suites after last commit
jkbradley
committed
fixes after rebasing on master
jkbradley
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[SPARK-4789] [SPARK-4942] [SPARK-5031] [mllib] Standardize ML Prediction APIs#3637

Commits on Feb 5, 2015

Added lots of classes for new ML API:

fixed compilation issues, but have not added tests yet

removing everything except for simple class hierarchy for classification

Added LinearRegression and Regressor back from ml-api branch

Modified ParamMap to sort parameters in toString. Cleaned up classes in class hierarchy, before implementing tests and examples.

Fixed bug from last commit (sorting paramMap by parameter names in toString). Fixed bug in persisting logreg data. Added threshold_internal to logreg for faster test-time prediction (avoiding map l…

Fixed issue with logreg threshold being set correctly

Updated docs. Added LabeledPointSuite to spark.ml

* Changed semantics of Predictor.train() to merge the given paramMap with the embedded paramMap.

added train() to Predictor subclasses which does not take a ParamMap.

* fixed LinearRegression train() to use embedded paramMap

Added JavaLabeledPointSuite.java for spark.ml, and added constructor to LabeledPoint which defaults weight to 1.0

Added methods:

Added spark.ml LinearRegressionSuite

rat fix

* Made ProbabilisticClassificationModel into a subclass of ClassificationModel. Also introduced ProbabilisticClassifier.

Fixed bugs from previous commit.

small cleanup

Fixed Linear/Logistic RegressionSuites

Fixed bug in LogisticRegression (introduced in this PR). Fixed Java suites

added blanket mima exclude for ml package

Updates based on code review. Major ones are:

fixed after sql datatypes PR got merged

Added MIMA excludes for VectorUDT (now public), and added DeveloperApi annotation to it

fixed JavaLinearRegressionSuite.java Java sql api

Fixed issues after rebasing from master (after move from SchemaRDD to DataFrame)

fixed test suites after last commit

fixes after rebasing on master

Commits on Feb 6, 2015

Added JavaDeveloperApiExample.java and fixed other issues: Made developer API private[spark] for now. Added constructors Java can understand to specialized Param types.

Last edits based on code review. Small cleanups