• JDK 8 (set JAVA_HOME environment variable)
• Maven
• Node
• angular cli (version 8.3.18)
• Mongodb

To get started you can simply clone this repository using git:

git clone https://github.com/niloferbano/Flow-based-Spark-ML.git
cd Flow-based-Spark-ML

• Start the backend

Change the path depending on your installation path MAVEN_PATH in 
src/main/java/de/tum/spark/ml/codegenerator/MavenBuild.java file
cd Flow-based-Spark-ML/src/main/resources/frontend/sparkML-ui
npm install
cd Flow-based-Spark-ML 
mvn clean install
mvn spring-boot:run or java -jar target/your-file.jar or you can use
choice of your IDE to run the application.

• Start the frontend

Navigate to path : Flow-based-Spark-ML/src/main/resources/frontend/sparkML-ui
npm start
Browse to http://localhost:4200

• Go to http://localhost:4200
• There are four steps involved in creating a flow for application generation.
  1. Start
  2. FeatureExtraction/FeatureExtractionFromTextFile
  3. ML model of your choice(Decision Tree/ KMeans/ Collaborative filtering)
  4. Save Model
• After creating a flow by dragging and dropping on the pipeline panel. Click Generate Jar button to submit the job.
• You will get either successful message with path to the package/code created by the application or a failure message in case something goes wrong.

• Decision Tree use case: https://archive.ics.uci.edu/ml/machine-learning-databases/covtype/
• KMeans Clustering use case: https://kdd.ics.uci.edu/databases/kddcup99/kddcup99.html
• Collaborative Filtering use case: https://storage.googleapis.com/aas-data-sets/profiledata_06-May-2005.tar.gz