AqueductClient is a simple Python wrapper around Quantopian's Aqueduct API.
It lets you easily create Pipeline executions and load their results.
AqueductClient supports Python 2.7 and Python 3.4+.
$ pip install git+https://git@github.com/quantopian/aqueduct-client.gitTo use AqueductClient, you need a Quantopian API Key. Once you have it, there
are several ways to use it:
- Use a credentials file: create
~/.quantopian/credentials(Linux or OS X) or%UserProfile%\.quantopian\credentials(Windows) and put the following in it:
[default] API_KEY = your_api_key
- Use an environment variable: set
QUANTOPIAN_API_KEYto your API key. - Pass your API key directly into the
create_clientmethod (see below) using theapi_keykwarg.
Note: Fuller documentation will be coming soon.
To use AqueductClient, create an instance. In this case, we are loading credentials from disk or environment variable.
from aqueduct_client import create_client
client = create_client()To run a new pipeline execution, use submit_pipeline_execution. Required parameters are code (string), start_date and end_date (date-like strings, dates, or Pandas timestamps). Optional parameters are name (string), params (a dict of parameters to pass to your pipeline), and asset_identifier_format (which can be "symbol" (default), "sid", and "fsym_region_id"). submit_pipeline_execution returns an id, which you can pass to get_pipeline_execution to monitor this pipeline's execution status.
get_all_pipeline_executions and get_pipeline_execution(id) let you load existing pipelines. Each pipeline has a status field, which can be RUNNING, SUCCESS, or FAILED.
For a successful pipeline, get_pipeline_results_dataframe(id) loads that pipeline's results into a pandas DataFrame. For a failed pipeline, get_pipeline_execution_error(id) shows you the information about the error.