An AutoML framework for implementing automated machine learning on data streams architectures in production environments.
From pip
pip install -U automl-streams
or conda
:
conda install automl-streams
from skmultiflow.trees import HoeffdingTree
from skmultiflow.evaluation import EvaluatePrequential
from automlstreams.streams import KafkaStream
stream = KafkaStream(topic, bootstrap_servers=broker)
stream.prepare_for_use()
ht = HoeffdingTree()
evaluator = EvaluatePrequential(show_plot=True,
pretrain_size=200,
max_samples=3000)
evaluator.evaluate(stream=stream, model=[ht], model_names=['HT'])
More demonstrations available in the demos directory.
Create and activate a virtualenv
for the project:
$ virtualenv .venv
$ source .venv/bin/activate
Install the development
dependencies:
$ pip install -e .
Install the app in "development" mode:
$ python setup.py develop