├── .gitignore ├── LICENSE ├── README.md ├── pom.xml ├── project ├── build.properties └── plugins.sbt └── src ├── main └── scala │ └── org │ └── apache │ └── spark │ ├── ml │ └── feature │ │ └── InfoThSelector.scala │ └── mllib │ └── feature │ ├── FeatureSelectionUtils.scala │ ├── InfoThCriterion.scala │ ├── InfoThCriterionFactory.scala │ ├── InfoThSelector.scala │ ├── InfoTheory.scala │ └── SelectorModel.scala └── test ├── resources └── data │ ├── test_colon_s3.csv │ ├── test_leukemia_s3.csv │ ├── test_lung_s3.csv │ ├── test_lymphoma_s3.csv │ └── test_nci9_s3.csv └── scala └── org └── apache └── spark └── ml └── feature ├── ITSelectorSuite.scala └── TestHelper.scala /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/.gitignore -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/LICENSE -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/README.md -------------------------------------------------------------------------------- /pom.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/pom.xml -------------------------------------------------------------------------------- /project/build.properties: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/project/build.properties -------------------------------------------------------------------------------- /project/plugins.sbt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/project/plugins.sbt -------------------------------------------------------------------------------- /src/main/scala/org/apache/spark/ml/feature/InfoThSelector.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/src/main/scala/org/apache/spark/ml/feature/InfoThSelector.scala -------------------------------------------------------------------------------- /src/main/scala/org/apache/spark/mllib/feature/FeatureSelectionUtils.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/src/main/scala/org/apache/spark/mllib/feature/FeatureSelectionUtils.scala -------------------------------------------------------------------------------- /src/main/scala/org/apache/spark/mllib/feature/InfoThCriterion.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/src/main/scala/org/apache/spark/mllib/feature/InfoThCriterion.scala -------------------------------------------------------------------------------- /src/main/scala/org/apache/spark/mllib/feature/InfoThCriterionFactory.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/src/main/scala/org/apache/spark/mllib/feature/InfoThCriterionFactory.scala -------------------------------------------------------------------------------- /src/main/scala/org/apache/spark/mllib/feature/InfoThSelector.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/src/main/scala/org/apache/spark/mllib/feature/InfoThSelector.scala -------------------------------------------------------------------------------- /src/main/scala/org/apache/spark/mllib/feature/InfoTheory.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/src/main/scala/org/apache/spark/mllib/feature/InfoTheory.scala -------------------------------------------------------------------------------- /src/main/scala/org/apache/spark/mllib/feature/SelectorModel.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/src/main/scala/org/apache/spark/mllib/feature/SelectorModel.scala -------------------------------------------------------------------------------- /src/test/resources/data/test_colon_s3.csv: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/src/test/resources/data/test_colon_s3.csv -------------------------------------------------------------------------------- /src/test/resources/data/test_leukemia_s3.csv: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/src/test/resources/data/test_leukemia_s3.csv -------------------------------------------------------------------------------- /src/test/resources/data/test_lung_s3.csv: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/src/test/resources/data/test_lung_s3.csv -------------------------------------------------------------------------------- /src/test/resources/data/test_lymphoma_s3.csv: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/src/test/resources/data/test_lymphoma_s3.csv -------------------------------------------------------------------------------- /src/test/resources/data/test_nci9_s3.csv: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/src/test/resources/data/test_nci9_s3.csv -------------------------------------------------------------------------------- /src/test/scala/org/apache/spark/ml/feature/ITSelectorSuite.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/src/test/scala/org/apache/spark/ml/feature/ITSelectorSuite.scala -------------------------------------------------------------------------------- /src/test/scala/org/apache/spark/ml/feature/TestHelper.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sramirez/spark-infotheoretic-feature-selection/HEAD/src/test/scala/org/apache/spark/ml/feature/TestHelper.scala --------------------------------------------------------------------------------