Karma-Decision-Tree-Service

A RESTful service for Decision Tree implementation 'rpart' in R.

Setting up the service

• Compile and create WAR file:

  • Open Terminal and go to dt-service folder
  • Run mvn clean package

• Copy the WAR file from dt-service/target/ to webapps folder of Tomcat

• Start the Tomcat server

Running the service

• To train and create a Decision Tree model:

  • Send a POST request with payload as contents of the training data in CSV format to

    http://host-name:port/dt-service/api/dt/train

  • By default, the last column is assumed to contain the class labels

  • To specify column-number of class labels, add classColumnNumber parameter:

    http://host-name:port/dt-service/api/dt/train?classColumnNumber=column-number

  • The response will contain a summary of the training phase and the name of the created model at its bottom

  • Note down the model name (which starts with 'DT' and ends with 'model')

• To test a model:

  • Send a POST request with payload as contents of the testing data in CSV format to

    http://host-name:port/dt-service/api/dt/test?modelName=model-name

  • To specify column-number of class labels, add classColumnNumber parameter:

    http://host-name:port/dt-service/api/dt/test?modelName=model-name&classColumnNumber=column-number

  • By default, the output of the testing phase is a confusion matrix

  • To specify the output type of the testing phase, add outputType parameter:

    • http://host-name:port/dt-service/api/dt/test?modelName=model-name&outputType=confusion_matrix
    • http://host-name:port/dt-service/api/dt/test?modelName=model-name&outputType=predictions