ThingsBoard Documentation

Documentation for using ThingsBoard IoT Platform.


Trendz has built-in instruments for timeseries prediction in few clicks. All required work, like data filtering, normalization and model training performed in the background. You can enable prediction for any fields including calculated fields.

Simple Prediction


In this example we monitor vibration of pumps on water station. We want predict when vibration will rich critical levels so we can schedule preventive maintenance for the pump and avoid unplanned downtime.

Now Our Chart is ready for making forecast:


Forecast resource usage


In this example, we forecast how much energy would be consumed by teh building during next year. We have Buildings, Apartments and each Apartment has energy meter. We need to aggregate telemetry from sensors on building level and then build a forecast.


Prediction Models

Trendz implements few methods for timeseries prediction:

Compare prediction with historical data

If you want to see how trained model looks like compared to historical data you need to enable Show Historical part checkbox in field configuration. Once enabled, predicted dataset will contain historical part predicted with the same model.