In this guide:
- Overview
- Introduction
- PHASE 5: Create an application that uses custom ML model
- PHASE 6: Connect Watson Assistant with your ML model
- IBM Cloud account
- Create a free account www.bluemix.net
The purpose of this lab is to guide the user in the data science life cycle from raw data to an application with a machine learning model. All this using IBM Cloud tools.
Part 1: Data science and data analytics part using Watson Studio, Cloud Object Storage and Watson Machine Learning services. https://github.com/sandra-calvo/Lab1-data-science
Part 2: Application development section using Node-RED and Watson Assistant services to connect a web application with the machine learning model deployed in Part 1. https://github.com/sandra-calvo/Lab2-application-development
IBM Cloud is a suite of cloud computing services from IBM that offers both platform as a service (PaaS) and infrastructure as a service (IaaS).
Node-RED is a visual tool for wiring the internet of things - connecting hardware devices, APIs and online services in a new and interesting way. Node-RED provides a browser-based flow editor that makes it easy to wire together flows using the wide range nodes in the palette. Flows can be then deployed to the runtime in a single-click.
- In a browser navigate to https://bluemix.net
- Select 'LOG IN' then enter your log in information and press 'SIGN IN'. You should see your dashboard.
- Select the 'CATALOG' view.
- Locate the Node-RED started service and click on it.
- Enter a name for your application, as shown below (host will automatically be completed). The host name must be unique on IBM Cloud, so please choose a name with your company name or initials to try to make a unique name. Press 'CREATE'.
- Your application is now staging and will be up and running in a short while. Click 'OVERVIEW' to see information about your application. The application will be ready in a couple of minutes. If you want to check the progeress click on the LOGS icon on the left side menu.
Note: If you are using Lite accounts your application will be in an awake mode. That means that if after 10 days your application has not been used IBM will stop it.
- When fully staged, click on the Visit app link, next to the green or half green circle, this launches the Node-RED main page.
- Configure your Node-RED editor. In this section, you will set up a username and password to protect your flow.
- Write an username and a password of your choice and click 'Next'. Remember that it does not have to be related to your IBM Cloud ID.
Now click Go to your Node-RED flow editor to open the flow editor.
- When using Node-RED we build our apps using this graphical editor interface to wire together the blocks we need. We can simply drag and drop the blocks from the left menu into the workspace in the center of the screen and connect them to create a new flow.
We are going to add new nodes to the Node-RED palette directly from the Node-RED window. For this lab we need the following nodes:
- node-red-dashboard
- node-red-contrib-watson-machine-learning
In the Node-RED window click on the three lines on the top right corner and in the menu, click on the "Manage palette". This will open the node menu where you can add new nodes to your application.
You will see the nodes that are installed by default and if you go to the 'install' tab you can search for any node package and add it directly to your app.
Search for the dashboard nodes by writing 'dashboard'. This will return multiple node packages, you need to install the package 'node-red-dashboard'. Find it in the search results and click on install.
This will prompt a window to confirm the installation. Click on install and wait few minutes, the application may require a restart. Click "Done" to close the left side menu.
After few seconds you will see the new nodes in your Node-RED palette.
Remember to repeat this process for the Machine Learning nodes: node-red-contrib-watson-machine-learning.
In this section we will build a simple flow to represent the user interface that will interact with our ML model created in Watson Studio.
Copy the content of the ML_form_UI.json file. (Located in the Box folder https://ibm.box.com/v/workshop260918) Import the flow by simply clickcing on the 3 white lines on the top right corner of the Node-RED window. Import - Clipboard.
Paste the text you copied from the file.
This flow reads input data from the user and calls the ML model to give a prediction in the UI.
You will need to do some editing on the Watson Machine Learning Node. Go to IBM Cloud, (www.bluemix.net) click on your dashboard and open the Watson Machine Learning service.
Click on Service Credentials and View credentials. Copy the credentials.
Back in Node-REd, double click on the purple node, Watson machine learning node, and click on the pencil to add your credentials.
Add your Username, Password, Host, Instance ID and Access key.
Note that it is also possible to change the looks of your user interface in the dashboard tab.
Deploy your application changes from the Deploy button on the top right side of the screeen.
The dashboard nodes added an UI to our Node-RED application. It also possible to change the looks of your user interface in the dashboard tab.
To access the UI go to: http://yourAppName.eu-gb.mybluemix.net/ui - UK
Remember that if you are in US, Germany Sydney the addredd will look slightly different: http://yourAppName.mybluemix.net/ui - US South
http://yourAppName.eu-de.mybluemix.net/ui - Germany
http://yourAppName.au-syd.mybluemix.net/ui - Sydney
Fantastic! Your web app is ready. Now you can interact with the machine learning model you created from the webapp. 👍
With IBM Watson™ Assistant service you can build a solution that understands natural-language input and uses machine learning to respond to customers in a way that simulates a conversation between humans.
Go to your IBM Cloud account and open the catalog. Look for Watson Assistant service and click on it.
Choose the region and space where you want the service to be created. Your organization will be filled by default.
You don't need to change the name if you don't want to, just click on 'Create'.
Once the service is created click on 'Launch tool' to access it.
Click on Log in with IBM ID and you will automatically access the service. It uses your IBM Cloud ID and password.
In the home tab you have videos and tutorials on how to get started building dialoges. Feel free to explore them. Let's move to the Workspaces tab.
The natural-language processing happens inside a workspace, which is a container for all of the artifacts that define the conversation flow for an application.
You can create a workspace and start from scratch or import an existing conversation. Let's start by importing a conversation. Download the bot_conversation.json file located in the Box folder.
Click on the import icon shown in the image below.
When you import a workspace, you can choose to import only the intents and entities, which can be useful if you want to build a new dialog using the same training data. In this case we will import everything.
As you make changes to your dialog, you can test it at any time to see how it responds to input.
- From the Dialog tab, click the conversation buble icon.
- In the chat panel, type some text and then press Enter.
- Check the response to see if the dialog correctly interpreted your input and chose the right response.
The chat window indicates what intents and entities were recognized in the input. In the dialog editor pane, the currently active node is highlighted
Feel free to create new intents for your bot.
Since we will need your Watson Assistant credentials and your workspace ID in the next step, this is a good moment to save them.Go to the deploy tab in the Assistant window. There you will find your workspace ID, username and password. Copy the credentials and save them for later.
Back to Node-RED window
Copy the content of ML_bot_UI.json and import the flow to Node-RED, same way you did in Step 15. The file is located in the Box folder.
This is the flow we are importing:
Connect the output of the orange node called "ML input" to the input of the purple node from the previous flow. Once you do connect the dots your flow should look like this:
Double click on the conversation node to edit the node with your own credentials saved in the previous step. Add your username, passworkd and workspace id and click Done.
Click on the Deploy button to save the changes in your application.
Go back to the UI and talk with your bot! You can ask to connect/start the model and the prediction result will be shown in the gauge graph.
Remember, to go back to your web app (in UK region) http://yourAppName.eu-gb.mybluemix.net/ui - UK
At this moment the bot is very basic. It can tell you about IBM Cloud and Watson Studio and can connect with the ML model created in Lab 1. To connec with the model write: "Start the model".
Congrats! You finished the lab. 👏 Here, take a 🍭 and enjoy your awesomeness!