AI

DataForge AI offers you the ability to train multivariate time series anomaly detection models on your Zabbix data to recognize anomalies in your Zabbix data.

The DataForge AI workflow consist of three main parts:

a dataset,
a model,
an inference

These three parts are explained in the AI chapter of the user manual:

Dataset configs

Dataset configs determine how data is extracted from your Zabbix server into datasets. Each dataset is listed in the config that was used to create it. Datasets are used as training data for the AI model and as validation sets to evaluate model performance in model tests.

Model configs

Model configs define how a model should be trained and which dataset should be used.

Inferences

Inferences determine how a model should be deployed and where it’s reconstruction loss should be sent.