Filter blob storage data and send to Machine Learning

yjay 256 Reputation points
2021-03-30T19:54:12.8+00:00

Hi,

As a kind of continuation of my previous question: Send data from IoT central to Azure Machine Learning Resource
I am wondering how I can filter my data to send to Machine Learning. I have multiple devices that are exporting data from IoT Central to Azure Blob Storage, I am wondering how I can filter the data for each device so I can send to a Machine Learning model?

Additionally, do I need to create multiple ML models for each device or can I send all filtered data to the same model? (For ex. If I have 5 devices collecting temperature data and I want to predict if the device user has a cold, I want to use the same model that predicts colds but I want to predict separately for each user based on their data)

Thanks so much!

Azure Machine Learning
Azure Machine Learning
An Azure machine learning service for building and deploying models.
2,571 questions
Azure Blob Storage
Azure Blob Storage
An Azure service that stores unstructured data in the cloud as blobs.
2,436 questions
0 comments No comments
{count} votes

1 answer

Sort by: Most helpful
  1. AshokPeddakotla-MSFT 27,491 Reputation points
    2021-03-31T10:19:14.807+00:00

    @yjay

    I am wondering how I can filter my data to send to Machine Learning. I have multiple devices that are exporting data from IoT Central to Azure Blob Storage, I am wondering how I can filter the data for each device so I can send to a Machine Learning model?

    You can add filters to reduce the amount of data exported. There are different types of filter available for each data export type:

    To filter telemetry, you can:

    • Filter the exported stream to only contain telemetry from devices that match the device name, device ID, and device template filter condition.
    • Filter over capabilities: If you choose a telemetry item in the Name dropdown, the exported stream only contains telemetry that meets the filter condition. If you choose a device or cloud property item in the Name dropdown, the exported stream only contains telemetry from devices with properties matching the filter condition.
    • Message property filter: Devices that use the device SDKs can send message properties or application properties on each telemetry message. The properties are a bag of key-value pairs that tag the message with custom identifiers. To create a message property filter, enter the message property key you're looking for, and specify a condition. Only telemetry messages with properties that match the specified filter condition are exported. The following string comparison operators are supported: equals, does not equal, contains, does not contain, exists, does not exist. Learn more about application properties from IoT Hub docs.

    Additionally, do I need to create multiple ML models for each device or can I send all filtered data to the same model? (For ex. If I have 5 devices collecting temperature data and I want to predict if the device user has a cold, I want to use the same model that predicts colds but I want to predict separately for each user based on their data)

    I am enquiring at the moment and will update you with my findings.