Bubble plugin for classification and regression
Bubble is a leader in web app development with no code. Thanks to the Peltarion plugin for Bubble you will be able to power your apps with AI, without having to write a single line of code.
Follow the Create a no-code AI app tutorial to see how you can create a web app on Bubble that connects to a Peltarion model.
Install the Bubble plugin
The Peltarion plugin page on bubble.io will give you a quick description and let you install the plugin in one of your applications. You can also install the plugin from inside your application creation by clicking on the Plugins tab, then on Add plugin, and by searching for Peltarion.
Please Note: For Image, Text, and Tabular Classification or Regression models please select the 'Peltarion' plugin. The 'Peltarion (text similarity) should only be used for Text Similarity'
Create a workflow
Create a 'Workflow' in the Bubble Workflow editor. Then choose the correct element on your Bubble App that will trigger the prediction to your model on the Peltarion platform. This is typically a 'Submit' button.
In order to connect to a model deployed on Peltarion, you will need to make sure that you have created a deployment from an experiment and that the deployment is set to be Enabled.
To use this model, you will need 2 items that you can get from the API information:
The URL of the model, this will tell your bubble app where the model can be reached.
The Token of the model, a secret code that you need to provide or the model will not accept the request.
After adding your first action in the workflow, you will enter these details and select the Peltarion plugin from the 'Plugins' option.
You will also need to provide data from your bubble application to the Peltarion model to get predictions.
Keys: Every input feature your model needs have a name displayed in the Deployment view. In the Bubble workflow, you will provide this name as a key.
Values: This is the particular value you want to use as input. You can connect data from your application to provide it as input value to the model.
Types: This specifies the type of data for being sent, e.g., numeric, image, text, etc.
FAQ and Potential Errors
Expected a number, but got a string (original data: "")
This error could be caused by a couple of issues. Here are some steps to troubleshoot:
Confirm the 'Type' of your inputs is the correct data type that your model expects in the Peltarion platform. For example: Type 1 should be set to 'number' if you are sending a number to the Peltarion platform for prediction.
If sending numbers to the Peltarion platform, set the 'Content format' of the Input to 'Integer'
Confirm that your deployment is enabled in the Peltarion platform.
Batch / Mupltiple Predictions
Currently, the Bubble Connector only supports a signal API request at once. If you would like full API capabilities, please use the 'Bubble API connector' plugin.
Get Full Prediction Results
The Bubble Plugin currently only supports the top prediction result in the prediction response.