Simple workflow API for MOA.

Makes use of the reactive API that Java introduced with Java 9:

https://community.oracle.com/docs/DOC-1006738

The following operators are currently available:

Package: com.github.fracpete.moaflow.source

• InstanceSource - generates instances using a stream generator

Package: com.github.fracpete.moaflow.transformer

• EvaluatorClassifier - evaluates a classifier
• EvaluatorRegressor - evaluates a regressor
• InstanceFilter - for filtering a data stream
• TrainClassifier - for training a classifier
• TrainRegressor - for training a regressor

Package: com.github.fracpete.moaflow.sink

• Console - simply outputs any objects it receives via toString()
• DrawTable - draws a table
• MeasurementPlot - for plotting a single measurement
• MeasurementsToCSV - stores the measurements from an evaluation as CSV
• MeasurementTableSawPlot - plots realtime chart on web-browser using Jupyter Notebook and IJava
• OutputLearningCurve - outputs the whole learning curve of MOA in a table format
• WriteModel - stores a model on disk

• ClassificationCsvAndTrain -- evaluates a classifier, outputs the statistics in a CSV, also trains the model and stores it on disk
• ClassificationPlot -- evaluates a classifier and plots a statistic.
• RegressionCsvAndTrain -- filters the data, evaluates a regressor, outputs the statistics in a CSV, also trains the model and stores it on disk
• RegressionPlot -- evaluates a regressor and plots a statistic.
• RegressionConstructSaveLoadRun -- constructs, saves, loads and runs a flow that evaluates a regressor, outputs the statistics in a CSV, also trains the model and stores it on disk
• ClassificationVariousSchemaTablePlot -- uses different evaluation schemes, output a table of learning curves and plots a statistic with TableSaw (only works with IJava and Jupyter Notebook)

• moa-flow-core - core functionality
• moa-flow-examples - example code
• moa-flow-app - meta-module for generating binary .zip and .deb package

• compile

  mvn clean install -DskipTests=true
  

• build .deb package

  mvn clean install deb:package -DskipTests=true