Description
Streaming machine-learning models can be trained incrementally as new data arrives. Data scientists and developers can build classifiers, regressors, anomaly detectors, and evaluation pipelines without retraining on full batches each time. It is a scientific library for scripts and notebooks.