Summary

In this chapter, you completed a use case of anomaly detection for predictive maintenance. You gained a comprehensive toolkit for analyzing anomalies in IoT data, starting from a visual exploration and moving on to a control chart and an auto-regressive model. Finally, you learned how to trigger an action automatically via REST.

This example use case works as a reference for other anomaly detection use cases. Data standardization, looping over multiple frequency bands, building a control chart, training an auto-regressive model, calculating the first- and second-level alarms, and finally, triggering an action via REST are standard pieces of any anomaly detection application.

In the next chapter, we will continue working with a regression model for another common use case – demand prediction. Also, we will utilize the power of Spark in modeling.