Estimator microservice

A continuously trained streaming prediction engine

If you have streaming feature vectors (sensor data, web logs, streaming event database) with readily measurable or otherwise knowable variables, you can train Estimator streaming machine learning engine to predict the latent variable values — to allow your app to intelligently prepare for the likely occurrences.

With a feedback loop producing online training data based on the actualized values for the predictions made, Estimator makes auto-tuned predictions based on the potentially evolving, even rapidly changing, inter-variable dependencies, while being able to learn new classes of the property being predicted directly from the streaming data vectors.