Anomaly Detection and Prediction using AWS Lookout for EquipmentCreated an AWS Lookout for Equipment project and ingested sample historical sensor data. The historical sensor data and label data is stored in an AWS S3 bucket. During ingestion, Lookout for Equipment prepares the dataset so that it can be used for multiple algorithms at training time. After the time series data has been ingested and a Dataset has been created, the next step is to train an anomaly detection model. An anomaly detection model is trained using part of the ingested data and the the model is then evaluated on the remaining period of the ingested data. After the model has been trained, the evaluation results can be viewed. Lastly, the model is deployed by creating an inference scheduler. Once the scheduler is up and ready, it will wake up in the specified interval and for each row with a timestamp that falls between two inference executions, the prediction and the raw anomaly score is generated and can be viewed. NOTE: The below demo video is hosted on YouTube and incase you are unable to view this demo video then access has to be specifically provided.
|