Processing the model diagnostics
When you train an anomaly detection model with Amazon Lookout for Equipment, you can visualize the results obtained over an evaluation period. These results are available in the console and you can also query an API to integrate and further post-process these results for your own needs.
At inference time, the inference scheduler reads new data from an input location on Amazon S3 and outputs the model results in an output location. Each inference execution creates a new directory named after the timestamp at which the scheduler woke up and each directory contains a single file in JSON Lines format. In Chapter 11, Scheduling Regular Inferences, you learned how to locate, download, and interpret the results contained in these files.
In this section, you will use a CloudFormation template that will deploy a CloudWatch dashboard that you can use to visualize training and inference results from Amazon Lookout for Equipment. You will then see how you...