Metadata-Version: 2.1
Name: datarobot-mlops-stats-aggregator
Version: 9.0.0
Summary: datarobot-mlops-stats-aggregator library to compute statistics aggregations
Home-page: http://datarobot.com
Author: DataRobot
Author-email: info@datarobot.com
License: DataRobot Tool and Utility Agreement
Platform: UNKNOWN
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.7
Classifier: Programming Language :: Python :: 3.8
Classifier: Programming Language :: Python :: 3.9
Classifier: License :: Other/Proprietary License
Classifier: Operating System :: MacOS
Classifier: Operating System :: POSIX
Classifier: Operating System :: Unix
Requires-Python: >=2.7
Description-Content-Type: text/markdown
Requires-Dist: Cython
Requires-Dist: drfaster (==9.0.0)
Requires-Dist: numpy
Requires-Dist: pandas

# MLOps Stats Aggregation Library (Python)

## Usage

The Python library for MLOps stats aggregation has two entry points:

### `aggregate_stats`

The `aggregate_stats` function in
[`aggregation.py`](https://github.com/datarobot/mlops-stats-aggregator/blob/master/stats-aggr-python/datarobot/stats_aggr/aggregation.py)
is the primary entry point into the library. This function accepts dataframes of raw features and/or predictions and
aggregates statistics suitable to be submitted to DataRobot MLOps for deployment monitoring. In order to use this function,
the types of each tracked feature must be specified, and certain feature types (e.g. currency) require additional format
information. See the docstring and type declarations of `aggregate_stats` for details about specific requirements.

### `merge_stats`

The `merge_stats` function defined in
[`merging.py`](https://github.com/datarobot/mlops-stats-aggregator/blob/master/stats-aggr-python/datarobot/stats_aggr/merging.py)
is used to merge together the outputs of calls to `aggregate_stats`. It returns a value with the same shape as that
returned by `aggregate_stats`. Merging stats makes sense (for example) in the MLOps Tracking Agent, which will buffer
stats aggregated by the MLOps Reporting Library. The agent can merge those buffered stats before sending them in a
single payload to DataRobot MLOps.


