Skip to content

IBM Cloud Pak for Data: data collection

To develop the anomaly predictive service we first need to access the data. We have two datasources in this example: the product information and the telemetries data coming from the different Reefer Containers. With the telemetries we should be able to assess anomaly. The Telemetries are saved to a noSQL database. We are using MongoDB on IBM Cloud.

Using Mongo Compass, we can see one of telemetry document as saved into MongoDB.

Telemetries in Mongo Figure 1: Mongo DB Compass: ibmcloud.telemetries collection

It is important to note that the Json document has sensors document embedded. As we will see later they will be mapped to different tables in Cloud Pak Virtualization.

As part of the data governance capability, a user with data engineer role can do the following tasks:

  • Define one to many connections to the remote different data sources
  • Create virtual assets to materialize tables and views from the different data sources
  • Assign an asset to an exisint project or a data request (governance object to ask to access data)

Define connection

First we need to get the connection information for the MongoDB database. See this note for information about Mongo DB instance on IBM Cloud.

  • Get the information about the data connection.

Mongo connection Figure 2: Mongo DB on IBM Cloud connection information

  • Then download the TLS certificate as pem file:
ibmcloud login -a https://cloud.ibm.com -u passcode -p <somecode-you-get-from-your-login>
# Define your resource group
ibmcloud target -g gse-eda
ibmcloud cdb deployment-cacert gse-eda-mongodb > certs/mongodbca.pem

Back to Cloud pak for Data, an administrator may define connections as a reusable objects by entering the data sources information. The figure below illustrates the connection configuration to the Mongo DB running on IBM Cloud:

CP4D add connection Figure 3: Define connection in CP4D

Add connection in Cloud Pak for Data

Virtualization may help automatically group tables without moving data, so we can group different data elements into a single schema.

Create a new project

Once logged into Cloud Pak for Data, create a new project. A project is a collection of assets the analytics team work on. It can include data assets and Notebooks, RStudio files Models, scripts...

From the main page select the project view:

Analytic project Figure 4: Top level navigation menu

and then new project, and select analytics:

Add project

Figure 5: Add project

Select an empty project:

Select project type Figure 6: Select project type

Enter basic information about your project

Project metadata Figure 7: Project metadata

The result is the access to the main project page:

Project main page Figure 8: Project main page

Now we need to define data assets into our project...

Data Virtualization

As introduced in this paragraph, we want to use data virtualization to access the historical telemetry records: The data engineer uses the Data virtualization capability to search for existing tables and add the tables he wants in the cart. For that, he uses the Virtualize menu

Data Virtualization menu Figure 9: Data Virtualization menu

and then selects Mongo DB in the Filters column and may be apply some search on specific database name.

Figure 10: Data Virtualization on Mongo DB

Once done, he selects the expected tables and then use Add to cart link. It is important to note that we have two tables to match the telemetry json document and the sensors sub json document.

The next step is to assign them to a project:

Figure 11: Data Virtualization cart and tables

Create a joined view

We need to join the telemetries and the sensors data into the same table, to flatten the records. In the current Mongo document, there is a 1 to 1 relationship between telemetry and telemetry sensor, so it is easy to flatten the model in one table.

In the Data Virtualization, as a data steward, we select My Virtualized data, and then select TELEMETRICS and TELEMETRICS_SENSORS tables, then the Join view. Within this new panel, we create a join key, by dragging the TELEMETRICS_ID and _ID together:

Joining tables Figure 12: Joining tables

Once joined, a new view is created:

Joined view Figure 13: Join view

We see, now, those new assets as part of the project (Use Add to project). The figure below show this new asset in the project:

Telemetries asset in the project Figure 14: Telemetries asset in the project

With some data: Telemetry Data Figure 15: Telemetry data

Note

It is important to note that building those views will create new connection to the database engine that can be accessed by using external tool.

Next

Next is to start working within a model → Next →