The application demonstrates usage of IBM Predictive Analytics Bluemix offering. This is also an extension of Big Data University Predicting Customer Satisfaction course; while participation in this course helps, it is not obligatory.
Application is based on Node.js and Express framework. It uses Predictive Analytics service API to integrate with IBM SPSS Modeler analytics.
Within this sample scoring application you are able to:
- select one of IBM SPSS Modeler streams uploaded to Predictive Analytics service on Bluemix
- specify which source node should be used as scoring input
- verify required input data schema (fields name and type)
- drag and drop csv file with input data for scoring (or double click on input data field to open file browser)
- call Predictive Analytics service scoring API using „Perform Calculations” button
- display scoring result in form of table
- IBM ID to login to Bluemix; see free trial article if you don't yet have it
- Cloud Foundry command line interface (only if you want to manually deploy to Bluemix)
- IBM SPSS Modeler (only if you want to modify the stream or create a new one; see details in SPSS Modeler stream preparation section)
- Node.js runtime (only if you want to modify the source code)
Whole preparation is described in details in IBM Predictive Analytics Service for Bluemix - General document, below steps are rather concise. In case of doubts refer to the document mentioned.
- From Bluemix catalog choose Predictive Analytics service. This service will later be binded with a Node.js application created from this sample. From this point note that the service itself offers a set of samples (this particular one among them) which can be automatically deployed and binded, which is the simplest way to see the sample in action.
- Upload the SPSS Modeler stream to your instance of Predictive Analytics service. This sample comes with SPSS Modeler stream (stream/customer-satisfaction-prediction.str) which can be used for that.
For a fast start, you can deploy the pre-built app to Bluemix by clicking the button
Note the application is fully functional only if binded with an instance of Predictive Analytics service, which need to be done manually. Check instructions how to do it.
As an alternative to the button, the application can be manually deployed to Bluemix by pushing it with Cloud Foundry commands, as described in next section. Manual deployment is also required when you want to deploy modified source code. Manual deployment consists of pushing the application to Bluemix followed with binding Predictive Analytics service to deployed application.
To push an application to Bluemix, open a shell, change to directory of your application and execute:
cf api <region>where <region> part may be https://api.ng.bluemix.net or https://api.eu-gb.bluemix.net depending on the Bluemix region you want to work with (US or Europe, respectively)cf loginwhich is interactive; provide all required datacf push <app-name>where <app-name> is the application name of your choice
cf push can also read the manifest file, see Cloud Foundry Documentation. If you decide to use manifest, you can hardcode the name of your instance of Predictive Analytics service instead of binding it manually, see services section manifest.yml.template file.
If this is your first Bluemix Node.js application, refer documentation of node-helloworld project to gain general experience.
See instructions
Running the application locally is useful when you want to test your changes before deploying them to Bluemix. To see how to work with source code, see Source code changes.
When the changes are ready, open a shell, change directory to your cloned repository and execute npm start to start the application. The running application is available in a browser at http://localhost:6001 url.
Application run locally can also use Bluemix Predictive Analytics service, see instructions how to link it.
The repository comes with pre-build app. If you want to rebuild application after modifying the sources:
- Follow steps listed in Requirements section
- Change to directory with downloaded source code or cloned git repo
- Execute
npm install - Execute
./node_modules/.bin/webpack
To empower any application with IBM SPSS Modeler analytics use Predictive Analytics service API.
The source code placed in pm_client.js file is an example of how to call this service API through JavaScript code. It demonstrates following aspects:
- Retrieval of all currently deployed models
- Metadata retrieval for a chosen predictive model
- Scoring with a chosen predictive model
As stated in Requirements section, from Bluemix catalog order an instance of Predictive Analytics service if you don't yet have it. Next step is to connect your deployed application with service, which is called binding. There are a few options to achieve that in Bluemix environment, link describes binding either by Bluemix user interface or by using cf cli.
- Deploy application to Bluemix and bind it to Predictive Analytics service.
- Go to the application overview pane, choose binded Predictive Analytics service and press 'Show Credentials'. Copy the pm-20 credentials json part (url, access_key).
- Create ./config/local.json file by copying ./config/local.json.template file. Edit the local.json file and paste obtained pm-20 credentials.
- Start your local application. You should be able to interact with the Predictive Analytics service e.g. by listing the models uploaded.
With this sample application IBM SPSS Modeler stream file customer-satisfaction-prediction.str is included. The stream file is located under stream subdirectory. The scoring branch (highlighted in green) is trained predictive model (Support Vector Machine with RBF kernel).
The stream requires input data on the source node when making a scoring request, as shown below. The sample input data scoreInput.csv file can be found under data subdirectory.
The result of scoring request is represented by 'Output data' Table node. The predicted churn and probability of churn are represented by 'Predicted Churn' and "Probability of Churn" fields defined in Table node.
The code is available under the Apache License, Version 2.0.