Offline RCA Bot

Click on vuCoreML on the left tab and navigate to the Workspaces section.

The workspaces page shows a list of previously configured Workspaces. Click on the + icon to create a new Workspace.

You can now configure the workspace; the workspace comprises 5 major sections

  • Details
  • Schema
  • Signalizers
  • Bot Settings
  • Storyboard 


Enter the Workspace Name, Description, and choose the Category as RCA, and choose the Run Type as Offline.

Note: Choose Run Type as Online for live data and Offline for third-party systems or data in CSV files. For the Offline data use the data imported using Import Data.

Click on Create to create your Workspace.


Once Workspace is created, you will be directed to the Schema page, where you can configure the schema which comprises the Journey, Components, and Graphs section. The schema is the place where you’ll have to define the business journey and its metrics. 

Metrics can be categorized into three types,

  • Lead Indicators (Business impacting metrics)
  • Operational Indicators (Application/Infra/Underlying metrics)
  • External Indicators (External metrics like holiday or maintenance window)

Note: You must categorize the metrics accurately because the incidents will be detected primarily based on the lead indicators. 


The journey will be the super-set of all metrics and components (i.e. you can think of this as a business journey). You can categorize the metrics at the journey level if they don’t specifically come under any particular component. 

You can now click Journey, and add a new signal.

For each signal you will be adding, you’ll have to specify the data model and metric column in that model for this signal. Only metric columns in a data model are eligible to be indicators. 

You can use the listing option to specify the data model and respective metric column and then categorize that metric using the category listing option. Similarly, you can add other signals.


Usually, in a business journey, most of the metrics will be defined at the touchpoints/component level. In such cases, you can use the components section to categorize the metrics. You can now add a new component.

Then you can specify the component name. A Component can have two constituents.

  1. Signals as a Constituent: Signals/Metrics will be present inside this component 
  2. Components as a Constituent: Only components will be present inside this. Note that only those components that are already created before this step can be added here.

1. Signals as a Constituent

In the case of signals as constituents, you can now add a signal and follow the same procedure as you did for Journey signals to Categorize them.

Note: A particular signal (i.e., a unique metric for a data model) can be defined only once in the whole schema. It can either be defined as the journey or inside a component

2. Components as Constituent

For Components as constituents, you can add a new component and select the component as a constituent. 

Then in the components listing, all the components that we previously created will be shown, you can select them based on the requirement and save them.


The Graphs are another important part of the VDS Schema. It is used to define the topology of the business journey/system after metrics categorization. Graphs can be specified at two levels, 

  • A journey-level graph (A graph defining all the components in a business journey)
  • A component-level graph (A graph defining all the components inside a specific component. This can be used for components with constituents as components) 

You can click Add New Graph, and select the graph type:

After selecting the graph type, you can create connections. Each connection acts like a link between two touchpoints/components. For each connection, you’ll have to specify the following 

  • From – The component where it is coming from 
  • To – The component where it is going to 
  • Direction – It could be UniDirectional/BiDirectional
  • Strength – The strength talks about how strong the relationship between two components i.e how much impact would a problem in one component affect the other. The listing contains High/Medium/Low.

And save the connection for that graph type. Similarly, if you want to create a connection for a particular component, you can follow the same approach.

Click on Submit Schema to get the Schema running.


After successfully submitting the Schema, you will be directed to the Signalizers page. The signalizers page contains the list of metrics configured in the Schema page along with information on ML techniques that will be running for the respective metrics 

We have a time filter next to the activate button to fetch the data from.

If you want to change the hyper-parameters for a particular metrics ML method, advanced users or ML Engineers can click the edit button of the metrics ML method.

It will direct you to the hyper-parameter editing page. On completion of editing, you can click the update for it to override the default parameters.

Now you can either globally activate the signalizers or activate only specific metric signalizers as per requirement locally (at the action section of each metric listed on this page) 

After activation, a pop-up will come up where you can click the Activate button.

After clicking, the vuRCA Bot will start creating the required pipelines. Once the pipelines are created, all the signals configured in the workspace activation buttons will be switched on.

You can click the right arrow to go to the next section >> Bot Settings.

Note: In the offline RCA, you can’t manually deactivate the Signalizer. Once the job is done, it will automatically get turned off.

Bot Settings

You can specify the Topological Correlation Frequency (in minutes) and Training Frequency (in Days). 

 Note: The Topological Correlation Frequency should be a value that reflects how frequently you want to monitor real incidents and find the root cause. The Training Frequency should be the value that reflects how often you want to train the model with fresh data.

Again, you need to activate the pop-up.

If RCA Bot is successfully configured, you’ll get the activation message.

After this step, you can start expecting incidents on the incidents page if any real incidents are occurring in the system/journey for respective times. 


The storyboard contains insights into a workspace. The initial section gives an overview of the list of metrics configured and their roles and health. 

Note: The Storyboards can be viewed in a separate browser window for ease of navigation and better user experience. To do this, users need to click the Open New Window button which will land the respective storyboard group in a separate window.

If you want to get insights on configured ML methods for these metrics, you can further select the CHI Storyboard and Anomaly Storyboard as follows.

Further Reading:

  1. RCA Bot
  2. Time Series Analysis
  3. Event Correlation


Browse through our resources to learn how you can accelerate digital transformation within your organisation.