Architecture of Catalyst Zia Services

A generic, end-to-end workflow of the data analysis and prediction process by various Zia components is described below :

1. When you access any Zia component from the Catalyst console, APIs, or SDKs, an API call will be triggered to the Catalyst servers, along with the input data provided by your application. Pre-trained machine learning models are pre-loaded and stored within Catalyst’s servers. Based on the specific Zia component being accessed, corresponding ML models will be used for prediction.

2. The input data will then undergo pre-processing to be converted to formats recognizable by the corresponding ML model.

3. The pre-processed data is then sent for prediction to the Zia ML models, and based on the evaluation metrics specified, predicted results are returned as the response. The prediction mechanism also depends on the accuracy and volume of the data being fed.