Changes for the upcoming release can be found in the "changelog" directory in our repository.
- Add
RelativeTimeEncoder
to calculate the relative time based on a timestamp column. #67
- Added missing examples in
mlnext.score
and updated other examples. #54 - Dropped support for python 3.8 and added support for 3.13. #65
- Added FeatureCreator. #63
- Added
LengthTransformer
to pad or truncate inputs to a specific length. #64
- Removed argument
y_true
frommlnext.score.pr_curve
. Usey
instead. #41 - Removed
mlnext.score.eval_sigmoid
. Usemlnext.score.apply_threshold
instead. #42
- Added missing examples in
mlnext.score
and updated other examples. #54 - Dropped support for python 3.8 and added support for 3.13. #65
- Added FeatureCreator. #63
- Added
LengthTransformer
to pad or truncate inputs to a specific length. #64
- Removed argument
y_true
frommlnext.score.pr_curve
. Usey
instead. #41 - Removed
mlnext.score.eval_sigmoid
. Usemlnext.score.apply_threshold
instead. #42
- Parameter
value
is now optional inmlnext.pipeline.Fill
to allow both filling methods. #46 - Fixed documentation of
mlnext.pipeline
. #47 - Allow
ColumnSelector
to be fitted on data ifkeys
is None. #50 - Fixed type of parameter
mapper
inColumnRename
. #51 - Changed build system to hatchling and updated dev internals. #58
- Added
mlnext.utils.convert_sequences
which converts sequences inside a mapping to a mapping by the index. #48 - Added
mlnext.anomaly.hit_rate_at_p
. #53 - Add py.typed to indicate mypy/typing support. #56
- Added parameters
reduce
andmask
tomlnext.rank_features
. #57
- Marked
mlnext.save_config
for removal in 0.7. #59 - Deprecated parameter
keys
in favor ofcolumns
inColumnSelector
. #60
- fixed
mlnext.find_anomalies
for scalar inputs (every dimension 1) #52
- Updated the signature and docstrings of
mlnext.plot.plot_signals
. #32 - Refactored
mlnext.score.eval_sigmoid
. #33 - Removed
roc_auc
frommlnext.score.eval_metrics
. #35 - Fixed inconsistent
typing
import. #38 - Removed
neg_label
fromscore.apply_threshold
. #40
- Added
mlnext.pipeline.ClippingMinMaxScaler
. #37 - Added
mlnext.recall_anomalies
and integrations formlnext.eval_metrics
,mlnext.pr_curve
. #39
- Deprecated
y_true
in favor ofy
inmlnext.pr_curve
. #41 score.apply_threshold
superceedes the functionality ofscore.eval_sigmoid
and is removed in 0.6. #42
- Fixed package name in readme. #31
- Changed mlnext.score.apply_threshold to being inclusive for the positive class. #24
- Cleaned
mlnext
namespace. #28 - Added parameter
k
tomlnext.anomaly.apply_point_adjust
from https://arxiv.org/abs/2109.05257. #29 - Remove size check of other dimensions in
mlnext.utils.truncate
. #30
- Added
mlnext.score.pr_curve
. #25 - Added adaptive
marker_size
tomlnext.plot_error
. #26 - Added
stride
tomlnext.temporalize
andmlnext.detemporalize
. #27
- Removed
mlnext.io.load_model
andTensorflow
dependency. #10 - Changed the roles of
x
andx_pred
inplot_signals
. #12 - Fixed typing of
np.ndarray
. #20 - Refactored
mlnext.io
. #22
- Added legend to
plot_signal
. #11 - Added
norm_log_likelihood
andbern_log_likelihood
. #13 - Added
mlnext.anomaly
. #14 - Functions in
mlnext.plot
now optionally return the figure withreturn_fig
. #15 - Added
mlnext.score.kl_divergence
for two normal distributions. #16 - Added example images for
mlnext.plot
. #17 - Added
mlnext.anomaly.apply_point_adjust_score
. #18 - Added MIT license with PLCnext Technology Copyright. #19
- Added truncation and shape assertion methods in
mlnext.utils
. #21 - Added
mlnext.utils.rename_keys
andmlnext.utils.flatten
. #23
- Added Digital Factory now introduction and legal notice to documentation. #7
- Added gradient based feature augmentation. #8
- Fixed installation of package. #5
Initial Release.