You have completed this hands-on module to upgrade your smart thermostat application to take advantage of a trained machine learning model. You have now previewed the end-to-end experience of how to build an IoT application: acquire physical data from sensors, synchronize state between the edge and the cloud, build a serverless application to control an edge device, and transform data in to actionable insights with machine learning.
If you are looking for ideas on how to extend the solution, here are a few:
Between this solution and the previous one (Smart thermostat), you created the following resources in AWS:
Your account will continue to accrue metered charges in three ways. First, for provisioned compute resources like the SageMaker endpoint that is hosted on a virtual instance. These resources have hourly charges. Second, for storage of data in resources like S3 buckets and your IoT Analytics project (channel, data store, and processed data set results). These resources have billing dimensions like Gigabyte-month. Third, for event-driven activity like publishing messages to IoT Core, invoking the Lambda function, and processing events in IoT Events, you are billed per event. So if you were to turn off your device, the event-driven activity would cease. You should destroy any of these AWS resources if you are done with the solution and don’t intend to continue using them.
The SageMaker endpoint is the most expensive resource as it it has an hourly charge. We recommend that you destroy this resource when you are done using it. Your model itself will persist and could be redeployed later. To destroy the SageMaker endpoint:
To destroy IoT Analytics storage resources:
The IoT Core rule, Lambda function, IoT Events input and detector, and IoT Analytics pipeline only incur further charges as you use them. If you no longer need them, you can destroy each by deleting them from their respective resource detail pages in their respective management consoles:
Go to the next tutorial, Intro to Alexa for IoT.