Canvas overview

Canvas is a workspace where you can create flows or machine learning models by importing datasets and adding recipes. After you create a project, you will be navigated to the Canvas where you can create, test, and train models using various recipe types. There are predefined templates available for data cleaning, data preparation, data analysis, feature engineering, model building, model prediction, and visualization. Specifically, the AI-assisted recipe enables you to provide the text prompt to generate the code recipe in seconds and use it in the pipeline. You can use a rapid canvas recipe to run end-to-end ML flow automatically by uploading a dataset.

You can perform the following tasks on the canvas to create flows:

  • Input the dataset

  • Execute or run recipes (Template, AI-assisted or rapid model)

  • Get the output dataset or a chart, or model

Getting familiar with these areas will help business users to build models more efficiently without writing any Python code or ML expertise. With a user-friendly interface, any user can get started with building machine learning models.

Building blocks and significance

The following are the building blocks you view while building the data pipeline.

Building blocks used in the flow

Icon

Name

Description

../_images/dataset_icon.png

Dataset

This icon represents a dataset. It is displayed when you import a dataset onto the canvas successfully, or when a dataset is generated after running a recipe.

../_images/recipe_icon.png

Recipe

This icon represents a recipe. It executes various transforms related to data cleaning, preparation, analysis, feature engineering, model building, prediction and visualization.

../_images/dashboard_icon.png

Dashboard

This icon represents a dashboard. You can visualize the data presented in the form of charts and scatter plots.

../_images/unbuilt_icon.png

Unbuilt recipe

This icon represents a recipe added to the flow but has no transformations added.

../_images/reciperunning_icon.png

Running the recipe

This icon represents the recipe execution in progress.

../_images/recipeerror_icon.png

Error

This icon represents a recipe error. It is displayed when the recipe is failed during its execution.

../_images/model_icon.png

Model

This icon represents a model that is generated after running a recipe.

../_images/artifact_icon.png

Artifact

This icon represents an artifact generated after running a recipe.

../_images/emptydataset_icon.png

Empty dataset

This icon represents an empty output dataset. It is displayed when the recipe run is in progress.

../_images/emptyrecipe_icon.png

Empty recipe

This icon represents a recipe added to the flow but is not executed.

../_images/apiconnector1.png

API Connector recipe

This icon represents API connector recipe added to the flow.

Dag view options

Use these Dag view options to change the view of the data pipeline.

Options on Dag

Icon

Name

Description

../_images/zoomin.svg

Zoom in

Use this option to enlarge the DAG.

../_images/zoomout.svg

Zoom out

Use this option to shrink the DAG.

../_images/fitview.svg

Fit view

Use this option to fit the Dag into the size of the screen.

../_images/autoalignment.svg

auto arrange canvas nodes

Use this option to auto arrange canvas nodes. This formats the building blocks and lines in the flow to make it easier to read. It also ensure that the connectors do not overlap with another.

../_images/curvedlines.svg

Curved connector

Use this option to use curved lines for aligning the building blocks.

../_images/straightlines.svg

Straight connector

Use this option to use straight lines for aligning the building blocks.

../_images/savecanvasnode.svg

Save canvas nodes orientation

Use this option to save the order in which you arranged the building blocks on the canvas manually.

../_images/expandall.svg

Expand all

Use this option to expand all the nodes in the data pipeline.

../_images/collapseall.svg

Collapse all

Use this option to collapse all the nodes in the data pipeline.

../_images/snippet.svg

Snippet Generator

Use this option to generate AI snippet for each node on the canvas.

Shortcut options on canvas

You can perform quick actions on different components on the canvas, such as dataset, artifact, recipe, model and chart, using the shortcut options. You must right-click on each component to view these options.

  • Dataset – View, Add AI-assisted recipe, Add Rapid Model recipe, AI Guide, and delete

  • Recipe – View, Run the recipe, check the logs, delete the recipe.

  • Chart – View and delete

  • Model – View, create prediction service, AI Guide, and delete.

  • Artifact – View and Delete.

Various options to create a machine learning flow on the canvas

Use these options to build machine learning flows.

How to access?

Click on the plus icon ico1 on the canvas to see these options under Datasets and Recipes:

Options on Dag

Option

Description

Dataset

The option to upload a dataset onto the canvas to do the predictions.

Artifact

The option to add artifacts onto the canvas.

Model

The option to add an existing model to the canvas.

Template

The option to add preexisting templates to the canvas to perform data cleaning, data pre-processing, feature engineering and model building.

AI-assisted

The option to use AI to generate the recipes or templates you want by providing a text prompt.

Rapid Model

The option to build machine learning models automatically by uploading the dataset and selecting the problem type.

AI Connector Recipe

The option to write Python code to generate datasets and run recipes.

The other options you can view on the canvas are:

  • Run option to run the recipes in the canvas. This option will remain disabled until the recipes are added to the canvas.

  • RapidCanvas AI guide to generate the steps required to develop a model for the selected use case.

  • Click the drop-down to switch between scenarios within the project.

  • Search for a specific entity in the canvas, using the Search entities option.

AI Guide

The AI guide feature is available for datasets, model, charts, and rapid canvas recipe.

Datasets and Models

To access the AI Guide for datasets or models, right-click on the respective dataset or model block. This redirects you to the AI Guide page. This will initiate a chat interface where you can interact with the AI specifically regarding the dataset or model you have selected. You can also ask a follow-up question on the response received.

Accessing AI guide from a dataset

../_images/aiguidedataset.gif

Charts

For charts, the process is slightly different. First, click on the chart block on the canvas. This action will open the charts page, where each chart has an AI Guide option. Clicking on the AI Guide for a particular chart will allow you to chat with the AI, asking questions and receiving guidance pertinent to the chart data.

Using the AI Guide

Once you have accessed the AI Guide for a dataset, model, or chart, you can engage in an interactive chat with the AI. Users can pose questions about the data within the selected component. For example, if you are working with a dataset, you can ask about missing values, or specific data points. When working with models, you can inquire about model performance metrics, training data specifics, or get explanations on how certain predictions are made.

For charts, you can ask for insights derived from the visual data, such as trends, outliers, or comparisons.

See also

To learn more about the other tabs on canvas, read the following sections: