Making Predictions via Hoeffding Classifier Model

Purpose:

This application demonstrates how to train a Hoeffding Classifier and to predict the sweet category from the sweet production stream in streaming manner.

Prerequisites:

  1. Download siddhi-gpl-execution-streamingml-x.x.x.jar from the following http://maven.wso2.org/nexus/content/repositories/wso2gpl/org/wso2/extension/siddhi/gpl/execution/streamingml/siddhi-gpl-execution-streamingml/ and copy the jar to {WSO2SIHome}/lib.
  2. Save this sample.
  3. If there is no syntax error, the following message is shown on the console: 
    * Siddhi App streaming-hoeffding-classifier-sample successfully deployed.

Executing the Sample:

  1. Start the Siddhi application by clicking on 'Run'.

  2. If the Siddhi application starts successfully, the following messages would be shown on the console.

    * streaming-hoeffding-classifier-sample.siddhi - Started Successfully!

Notes:

If you edit this application while it's running, stop the application -> Save -> Start. * Stop this Siddhi application (Click 'Run' on menu bar -> 'Stop'). * Start the application and check whether the specified events from the jms provider appear on the console.

Testing the Sample:

Note: The Hoeffding Classifier for streaming machine learning needs to be trained prior to perform prediction.

Training phase

Send events through one or more of the following methods.

Send events to ProductionTrainStream, via event simulator:
  1. Open the event simulator by clicking on the second icon or pressing Ctrl+Shift+I.
  2. In the Single Simulation tab of the panel, specify the values as follows:
    • Siddhi App Name: streaming-hoeffding-classifier-sample
    • Stream Name: ProductionTrainStream

  3. In the name and amount fields, enter the following and then click Send to send the event. 

    density: 50.4
solubility: 30.03
sweetType: candy

  4. Send more events including upto 3 unique sweet types as specified as the parameter during the training phase. 

    @info(name = 'query-train')
from ProductionTrainStream#streamingml:updateHoeffdingTree('classifierModel', 3, density, solubility, sweetType )

Send events to the simulator http endpoint through the curl command:
  1. Open a new terminal and issue the following command: 
    * curl -X POST -d '{"streamName": "ProductionTrainStream", "siddhiAppName": "streaming-hoeffding-classifier-sample","data": [50.4, 30.03, candy]}' http://localhost:9390/simulation/single -H 'content-type: text/plain'
  2. If there is no error, the following messages are shown on the terminal: 
    {"status":"OK","message":"Single Event simulation started successfully"}
Publish events with Postman:
  1. Install 'Postman' application from Chrome web store.
  2. Launch the application.
  3. Make a 'Post' request to the 'http://localhost:9390/simulation/single' endpoint. Set the Content-Type to 'text/plain' and set the request body in text as follows: 
    {"streamName": "ProductionTrainStream", "siddhiAppName": "streaming-hoeffding-classifier-sample","data": [50.4, 30.03, candy]}
  4. Click 'send'. If there is no error, the following messages are shown on the console: 
    "status": "OK",
"message": "Single Event simulation started successfully"

Testing phase

Send events through one or more of the following methods.

You may send events to ProductionInputStream, via event simulator
  1. Open the event simulator by clicking on the second icon or pressing Ctrl+Shift+I.
  2. In the Single Simulation tab of the panel, specify the values as follows:
    • Siddhi App Name: streaming-hoeffding-classifier-sample
    • Stream Name: SweetProductionStream
  3. In the name and amount fields, enter the following and then click Send to send the event. 
    density: 30.4
emperature: 20.5
Send events to the simulator http endpoint through the curl command:
  1. Open a new terminal and issue the following command: 
    curl -X POST -d '{"streamName": "SweetProductionStream", "siddhiAppName": "streaming-hoeffding-classifier-sample","data": [30.4, 20.5]}' http://localhost:9390/simulation/single -H 'content-type: text/plain'
  2. If there is no error, the following messages are shown on the terminal: 
    {"status":"OK","message":"Single Event simulation started successfully"}
Publish events with Postman:
  1. Install 'Postman' application from Chrome web store.
  2. Launch the application.
  3. Make a 'Post' request to the 'http://localhost:9390/simulation/single' endpoint. Set the Content-Type to 'text/plain' and set the request body in text as follows: 
    {"streamName": "SweetProductionStream", "siddhiAppName": "streaming-hoeffding-classifier-sample","data": [30.4, 20.5]}
  4. Click 'send'. If there is no error, the following messages are shown on the console: 
    "status": "OK",
"message": "Single Event simulation started successfully"

Viewing the Results:

See the output on the terminal. 

INFO {io.siddhi.core.stream.output.sink.LogSink} - streaming-hoeffding-classifier-sample : PredictionStream : Event{timestamp=1513610806272, data=[30.4, 20.5, candy, 0.0], isExpired=false}

@App:name("streaming-hoeffding-classifier-sample")
@App:description('Train a streaming Hoeffding Classifier and to predict the type of sweet.')


define stream ProductionTrainStream (density double, solubility double, sweetType string );

define stream SweetProductionStream (density double, solubility double);

@sink(type='log')
define stream PredictionStream (density double, solubility double, prediction string, confidenceLevel double);

@info(name = 'query-train')
from ProductionTrainStream#streamingml:updateHoeffdingTree('classifierModel', 3, density, solubility, sweetType )
select *
insert into trainOutputStream;

@info(name = 'query-predict')
from SweetProductionStream#streamingml:hoeffdingTreeClassifier('classifierModel', density, solubility )
select density, solubility, prediction, confidenceLevel
insert into PredictionStream;
Top