What is RubiStudio

RubiStudio is a visual model designer for data scientists. It helps to build, train, test, deploy, and publish data models. These published data models can be reused whenever required.
This enables you to maintain multiple versions of models; as well as integrate Python and R based models.

Creating a Workbook

To create a workbook, follow the steps given below.

  1. On the home page, click the Create icon ().
  2. Hover over the rubistudio tile and click the Create Workbook button.


    Figure: Creating a Workbook
    Create Workbook page is displayed.
  3. Enter the Name for your workbook.
  4. Enter the name in the Project field to create a new project.

    (info)

    Notes:

    • The workbook is created under the entered project name.
    • You can also type the name of the existing project in the Project field to select the project.
    • Project name is mandatory.
  5. Enter Description for your workbook.
  6. Review all fields and click Create.

    Figure: Create Workbook
    The workbook is created, and workbook canvas is displayed.

Understanding the Workbook Canvas

The workbook canvas is the area where you can build algorithm flows. When you open a new workbook, the following icons and fields are displayed.


Figure: Workbook Canvas

The workbook screen has four panes, as given below.

  1. Task Pane: This pane displays the datasets and algorithms that can be used in the workbook.
  2. Function Pane: This pane displays the various functions that can be performed on the workbook.
  3. Workbook Canvas: This is the work area where you can drag-and-drop algorithms and datasets to create algorithm flows.
  4. VIEW LOG: This tab tracks the activities performed in the workbook and keeps a log (track) of it.

The table given below describes the fields and icons present on the Workbook Canvas.

Table: Description of Fields for Workbook Canvas

Icon/Field


Description

Task pane


It displays the list of datasets and algorithms that can be used in the workbook.

Name of the workbook

It displays the title of the workbook.

Zoom

It helps to magnify the canvas to view detailing or move further away to take an overview look of the canvas.

Minimap

It displays a scaled down version of the workbook.

Save

It saves your workbook.

Save As

It saves your workbook with a different name.

Run

It runs the created algorithm flow.

Hyperparameters

It is set of values of the algorithm properties at which the algorithm gives best results.

Snapshot

It takes the snapshot of the created algorithm flow. The snapshot can be restored later for version control.
For more information, refer to Using Snapshot for Version Control.

Workbook Canvas


It is the area to experiment with your datasets and algorithms. You can use the drag-and-drop method to insert datasets and algorithms in the workbook canvas.

VIEW LOG


It displays the logs of the algorithm flow that are run, along with the trace of algorithm properties.

Building Algorithm Flow in a Workbook Canvas

The workbook canvas is your experimental area.
You can create, run, and train your algorithm flow in workbook canvas using simple drag-and-drop method, to add datasets and algorithms.
The table given below provides major differences between workbook and workflow.

Workbook Canvas

Workflow Canvas

It is an experimental area where you train your datasets and algorithms.

It is the production area where trained datasets are used to publish your algorithms as models.

It also contains classification and regression algorithms, along with other algorithms.

It contains all algorithms present in the workbook, except classification and regression algorithms.

You cannot schedule algorithms to run later.

You can schedule your final algorithms to run automatically at specific time intervals.

To use canvas, follow the steps given below.

(info)

Notes:

  • It is recommended to save your Workbook after every step. This ensures all the changes are saved.
  • To save the workbook, click Save icon ( ).
  1. Open the Workspace where you want to create your workbook. Refer to Changing Workspace.
  2. Create a Workbook. Refer to Creating a Workbook.
  3. Use drag-and-drop method to insert readers and algorithms from the Task Pane into the canvas.

    (info) Note:

    You can also double-click on the reader or algorithm to add it into the canvas.


    Figure: Adding Datasets & Algorithms in the Canvas
  4. Click the reader/algorithm and click the Node Connector, and then click the other reader/algorithm that you want to join with it.
    Similarly join the other elements.

    (info) Note:

    You can also join a single element to multiple elements if required.


    Figure: Joining the Elements
  5. Click the reader/algorithm and select the Properties displayed on the right-hand side.
    Here we have selected Data Fields (columns in your dataset) for the selected reader.

    Figure: Selecting Properties of Datasets and Algorithms
  6. Click the reader/algorithm and click the vertical ellipsis ( ). The functions available for the elements are as follows.

    Figure: Functions Present on Reader or Algorithm

    (info) Note:

    Publish as a model is available only for rubiML, Regression, and Classification algorithms.

    The table given below describes the functions present on the reader or algorithm.

    Table: Description of Functions for Elements

    Icon

    Icon Name

    Description

    Run

    It runs the selected node of the algorithm. Note: The predecessor node is expected to be successfully executed.

    Run from Node

    It runs all the nodes below, from the selected node.
    Note: All nodes before this node are expected to be successfully executed.

    Run till Node

    It runs all the nodes from the top node, until the selected node.

    Explore

    It lets you explore the result of the selected node.
    Note: You can explore only after the algorithm is successfully executed.

    Publish as a Model

    It publishes the algorithm flow as a model.
    Note: You can publish models only in rubiflow for Machine Learning algorithms.

    Delete

    It deletes the node.

  7. Click the Save icon ( ) on the function pane, to save the algorithm flow.

(info)Notes:

  • To run the algorithm, click Run icon ( ). For more information refer to Running a Workbook.
  • To stop the execution before completion, click the Abort icon ().
  • You can publish your models only if you are working in rubiflow. For publishing your models, refer to Publishing as a Model.

Working with Workbooks

Workbooks are the play area where you can train the algorithms before they can be used in production. There are mainly four functions that can be performed with workbooks.

  1. Opening a Workbook
  2. Editing a Workbook
  3. Running a Workbook
  4. Exploring a Workbook
  5. Deleting a Workbook

The following sections describe each of these functionalities in detail.

Searching a Workbook

You can search for a workbook by its name. It is especially helpful when the workbook list is long.

(info) Note:

Make sure you are in the correct Workspace, which includes the Workbook that you want to search.

To search a project, follow the steps given below.

  1. Open the Workspace that includes your project. Refer to Changing Workspace.
  2. On the home page, click Workbooks.
    Recent Workbooks for the selected workspace are displayed
  3. Type the workbook name in the Search field.
    As you start typing, the list of workbooks whose name or description match with the search string is populated, as shown in the figure below.

Figure: Searching a Workbook

(info)Notes:

  • Click the ellipsis corresponding to a project, to Run or Delete the workbook. Refer to respective topics for more information.
  • To open a workbook, hover over the workbook tile and click View.

Opening a Workbook

To open a workbook, follow the steps given below.

  1. On the home page, click Workbooks.
    Recent Workbooks for the selected workspace are displayed.
  2. Hover over the workbook you want to open and click View.

Figure: Opening a Workbook

Your selected workbook canvas is displayed.

Editing a Workbook

To edit a workbook, follow the steps given below.

  1. On the home page, click Workbooks.
    Recent Workbooks for the selected workspace are displayed.
  2. Hover over the workbook you want to edit and click View.
    The workbook canvas is displayed.
  3. Edit the workbook using the drag-drop method to insert datasets and algorithms from the task pane.
    For more information, refer to Building Algorithm Flow in a Workbook Canvas.
  4. Once the workbook editing is complete, click the Save ( ) icon on the function pane.
    Changes to the workbook are saved.

(info)Note:

To save the workbook with a different name, click the ellipsis ( ) on the function pane and then click Save As icon ( ).

Running a Workbook

You can run an existing workbook.
You can run a workbook either from Card View or from Workbook Canvas.

(info)Note:

After you have run the workbook, check the logs for more information about execution details. Refer to Viewing Logs in Workbook.

Running from Card View

To run a workbook from card view, follow the steps given below.

  1. On the home page, click Workbooks.
    Recent Workbooks for the selected workspace are displayed.
  2. Hover over the workbook you want to run, click the vertical ellipsis ( ), and then click Run.

    (info) Note:To stop the execution before completion, click the ellipsis and click Abort.

Figure: Running a Workbook

After the run is successfully completed, the status of the Workbook is displayed as Completed on the card.

Figure: Status of Workbook

The other status is displayed as,

  • Running: The run has initiated
  • Failed: The run has failed and could not be completed

Running from Workbook Canvas

  1. Open a Workbook. Refer to Searching a Workbook.
    Workbook canvas is displayed.
  2. Click the Run icon ( ) located in the top-right corner on the function pane.
    The workbook execution starts.

After the run is completed, the result is displayed with following icons:

— The algorithm flow has run successfully.

— The algorithm flow has failed to run.


(info) 

Notes:

  • To stop the execution before completion, click the Abort icon ( ).
  • To explore the result, click the element and click the vertical ellipsis ( ), and then click Explore.

Viewing Logs in Workbook

When you run an algorithm, rubiscape generates a log of the tasks that are performed in background. The log shows task name, type, duration, and status. The logs and activity trace are particularly helpful for troubleshooting.
To view log, follow the steps given below.

  1. Open a Workbook. Refer to Opening a Workbook.
  2. Click the Run icon ( ) located in the top-right corner on the function pane.
    The workbook execution starts.
  3. During the execution or after completion, Click VIEW LOG.
    Logs are displayed as shown in the figure below.

Figure: Workbook Log

Log: It displays the details such as Task Name, Task Type, Status, and so on.
Trace: It displays the details of execution. You can track the error if your execution fails.


Figure: Workbook Log Trace

The log trace shows more details such as the username, workbook name, and timestamp of each of the activities.
The log status can have one of these values: Yet to Start, Running, Completed, and Failed.

Deleting a Workbook

To delete a workbook, follow the steps given below.

  1. On the home page, click Workbooks.
    Recent Workbooks for the selected workspace are displayed.
  2. Hover over the workbook you want to delete, click the vertical ellipsis ( ), and then click Delete.

    Figure: Deleting a Workbook
    A message to confirm your action is displayed.
  3. To delete the workbook, click Delete.
    The selected workbook is deleted and a confirmation message is displayed.

Publishing Models

You can publish algorithms as models after their successful execution. A model can be reused in a workbook for training and experimenting, or can be used in a workflow for production.

(info) Notes:
  • This functionality is available only for rubiML Regression, rubiML Classification, AutoML, and rubitext Classification algorithms.
  • You can publish an algorithm only after it has run successfully.

To publish a model, follow the steps given below.

  1. Open the workbook that you want to publish as model. Refer to Opening a Workbook.
  2. Run the node that you want to publish. Refer to Running a Workbook.
  3. To publish the model, click the algorithm, click vertical ellipsis, and then click Publish as a Model icon ().
    Figure: Publishing Workbook as a model
    Your model is published. 
    The published models are available in rubistudio and rubiflow under Models. You can use them in your workbooks and workflows.




TABLE OF CONTENT