nannyml: post-deployment data science in python
APACHE-2.0 License
Bot releases are visible (Hide)
np.nan
to None
. (#387)
identifier
column from the documentation example for reconstruction error calculation with PCA. (#382)
Published by github-actions[bot] 8 months ago
incomplete
parameter in the SizeBasedChunker
and CountBasedChunker
keep
from the previous append
. This means that from now on, by default, you might have an additionalpyarrow
dependency to ^14.0.0
if you're running on Python 3.8 or up.Published by github-actions[bot] 8 months ago
Published by github-actions[bot] 8 months ago
confidence_deviation
properties in CBPE metrics (#357)
UnseenValuesCalculator
Published by github-actions[bot] 11 months ago
CalibratorFactory
to align with our other factory implementations. [(#341)]Calibrator
interface with *args
and **kwargs
for easier extension.ResultComparisonMixin
to allow easier extension.DatabaseWriter
support for results from MissingValuesCaclulator
and UnseenValuesCalculator
. SomeNaN
value. (#326)
specificity
calculation, both realized and estimated. Well spotted @nikml! (#334)
NaN
values. Major thanks to the eagle-eyed @giodavoli. (#333)
DatabaseWriter
. An inspiring commit that lead to some other changes.NaN
values when fitting univariate drift. [(#340)]Published by github-actions[bot] over 1 year ago
Published by github-actions[bot] over 1 year ago
nannyml.io
package, thanks @maciejbalawejder (#286)
numpy
to be <1.25
, since there seems to be a change in the roc_auc
calculation somehow (#301)
Ranker
implementations (#297)
mendable
in the docs (#295)
*args
and **kwargs
in Result.filter()
and subclasses (#298)
roc_auc
in CBPE
(#294)
BaseException
and not Exception
(#307)
Published by github-actions[bot] over 1 year ago
nannyml.runner
and the accompanying configuration format to improve flexibility (e.g. settingnannyml.runner
nannyml.io
interfaces, especially the nannyml.io.RawFilesWriter
nannyml.base.Result
az://
URLs in the CLI, thanks @michael-nml (#283)
Published by github-actions[bot] over 1 year ago
business_value
metric for both estimated and realized binary classification performance. It allowsPublished by github-actions[bot] over 1 year ago
treat_as_categorical
parameter to univariate drift calculator (#239)
Published by github-actions[bot] over 1 year ago
Published by github-actions[bot] over 1 year ago
plot()
function calls for data reconstruction results, univariate drift results,OrdinalEncoder
instead of LabelEncorder
in DLE. This allows us to deal with "unseen" values in they_pred
column as continuous values for the included sample binary classification data.Published by github-actions[bot] almost 2 years ago
nannyml.drift.ranker
module. The abstract base class and factory have been dropped in favorHellinger distance
, used for continuous variables.CorrelationRanker
ranks columns based onmetrics
parameter of the result.filter()
function, as per special request.Published by github-actions[bot] almost 2 years ago
mypy
to a new version, immediately resulting in some new checks that failed.Wasserstein distance
for continuous variables,L-Infinity distance
for categorical variables.mypy
issues concerning 'implicit optionals'.Published by github-actions[bot] almost 2 years ago
SizeBasedChunker
and CountBasedChunker
.incomplete
, that can be set to keep
, drop
or append
.nannyml.drift
module. The intermediate structural level (model_inputs
, model_outputs
, targets
)UnivariateDriftCalculator
. The old built-in statistics have beenMethods
, allowing us to add new methods to detect univariate drift.filter
method, which returns a new Result
instance, with a smaller 'scope'. Then turn thisResult
into a DataFrame using the to_df
method.nannyml.io
module with new Writer
implementations: DatabaseWriter
that exports data into multiplePickleFileWriter
which stores theResults
on local/remote/cloud disk.UnivariateDriftCalculator
.Published by github-actions[bot] about 2 years ago
dependencybot
dependency updatesstalebot
setupy_pred_proba
values to calculate realized performance. Fixed for both binary andPublished by github-actions[bot] about 2 years ago
timestamp_column_name
required by all calculators and estimators optional. The main consequences of thiss3fs
dependencyRunner
classPublished by github-actions[bot] about 2 years ago
problem_type
parameter to determine the correct graph to output when plotting model output driftproblem_type
argument in the Quickstart guide