How to resolve conflict between users decision and IoT system?

Question

In an Internet project, we have to use machine learning to resolve the conflict between the user's decisions and the system. In this project we use TICKstack (Telegraf, Influxdb, Chronograf, Kapacitor). To better understand the subject, consider the following two scenarios. Is there a plugin in Kapacitor to do this?

Conflict scenarios

Answer 1

I would recommend:

• store user actions in one time series
• store system predictions in another series
• compute the actuation signal based on those two, as a new series

The latter can be done with scripts in Kapacitor. Presumably user input should always override the "smart" system. You may need to consider user input for a boolean variable as tri-state: Unknown (don't care) | Explicitly On | Explicitly Off