An open-source, low-code machine learning library in Python
log_experiment
more configurable (https://github.com/pycaret/pycaret/pull/2334, https://github.com/pycaret/pycaret/pull/2335)return_train_score=False
use the old output format (https://github.com/pycaret/pycaret/pull/2333)dashboard_logger
key error during setup
(https://github.com/pycaret/pycaret/pull/2311)check_fairness
exception when index is not and ordinal number - thanks to @reza1615 (https://github.com/pycaret/pycaret/pull/2055)create_api
function (https://github.com/pycaret/pycaret/pull/2146)drift_report
can now work with unseen data - thanks to @reza1615 (https://github.com/pycaret/pycaret/pull/2183)return_train_score=True
). Passing an unseen dataset with the label column to predict_model
will now calculate the metrics for that dataset - thanks to @levelalphaone (https://github.com/pycaret/pycaret/pull/2237)numba<0.55
(https://github.com/pycaret/pycaret/pull/2056)create_app
(https://github.com/pycaret/pycaret/pull/2044)optimize_threshold
function (https://github.com/pycaret/pycaret/pull/2041)create_docker
(https://github.com/pycaret/pycaret/pull/2005)create_api
(https://github.com/pycaret/pycaret/pull/2000)check_fairness
(https://github.com/pycaret/pycaret/pull/1997)eda
(https://github.com/pycaret/pycaret/pull/1983)convert_model
(https://github.com/pycaret/pycaret/pull/1959)plot_model
(https://github.com/pycaret/pycaret/pull/1940)drift_report
functionality to predict_model
(https://github.com/pycaret/pycaret/pull/1935)dashboard
(https://github.com/pycaret/pycaret/pull/1925)grid_interval
parameter to optimize_threshold
- thanks to @wolfryu (https://github.com/pycaret/pycaret/pull/1938)tree
plot - thanks to @yamasakih (https://github.com/pycaret/pycaret/pull/1982)pyyaml<6.0.0
to fix issues with Google ColabFix_multicollinearity
would fail if the target was a float (https://github.com/pycaret/pycaret/pull/1640)interpret_model
not always respecting save_path
(https://github.com/pycaret/pycaret/pull/1707)compare_models
- thanks to @reza1615 (https://github.com/pycaret/pycaret/pull/1739)ignore_features
doesn't exist in the dataset - thanks to @reza1615 (https://github.com/pycaret/pycaret/pull/1793)probability_threshold
argument in various methods (https://github.com/pycaret/pycaret/pull/1858)stack_models
and calibrate_models
(https://github.com/pycaret/pycaret/pull/1849, https://github.com/pycaret/pycaret/pull/1858)RuntimeError
will now be raised if an incorrect version of scikit-learn
is installed (https://github.com/pycaret/pycaret/pull/1870)numba
(https://github.com/pycaret/pycaret/pull/1735)get_leaderboard
function for classification and regression modulesplot_model
and interpret_model
- thanks to @bhanuteja2001 (https://github.com/pycaret/pycaret/pull/1537)interpret_model
affecting plot_model
behavior - thanks to @naujgf (https://github.com/pycaret/pycaret/pull/1600)blend_models
and stack_models
throwing an exception when using custom estimators (https://github.com/pycaret/pycaret/pull/1500)errors="ignore"
parameter for compare_models
now correctly ignores errors during full fit (https://github.com/pycaret/pycaret/pull/1510)numba<0.54
(https://github.com/pycaret/pycaret/pull/1530)[full]
install by pinning interpret<=0.2.4
deploy_model()
with AWSTPESampler
options to improve convergence (in tune_model()
)interpret_model
- thanks to @IncubatorShokuhou (https://github.com/pycaret/pycaret/pull/1415)plot_model
under pycaret.classification
moduleremove_multicollinearity
considering categorical featuresgain
and lift
plots taking wrong arguments, creating misleading plotsinterpret_model
on LightGBM will now show a beeswarm plotpycaret.persistence
(https://github.com/pycaret/pycaret/pull/1324)remove_perfect_collinearity
option will now be show in the setup()
summary - thanks to @mjkanji (https://github.com/pycaret/pycaret/pull/1342)IterativeImputer
setting wrong float precisiontune_model
raising an exception when composed of listspycaret.clustering
- thanks to @susmitpy (https://github.com/pycaret/pycaret/pull/1372)address
in get_data
for alternative data sources - thanks to @IncubatorShokuhou (https://github.com/pycaret/pycaret/pull/1416)infer_signature
from MLflow logging when log_experiment=True
.finalize_model
when using GroupKFold CVmlxtend>=0.17.0
, imbalanced-learn==0.7.0
, and gensim<4.0.0
Modules Impacted: pycaret.classification
pycaret.regression
pycaret.clustering
pycaret.anomaly
pycaret.arules
pycaret.nlp
pycaret.regression
module. You can now generate interactive residual plots by using residuals_interactive
in the plot_model
function.display_format
is added in the plot_model
function. To render plot in streamlit app, set this to streamlit
.tune_model
in pycaret.classification
and pycaret.regression
is now compatible with custom models.pip install pycaret[full]
.raw_score
argument in the predict_model
function for pycaret.classification
module. When set to True, scores for each class will be returned separately.handle_unknown_categorical
is set to False in the setup
function, an exception will be raised during prediction if the data contains unknown levels in categorical features.predict_model
for multiclass classification now returns labels as an integer.pycaret. clustering
and pycaret. anomaly
pycaret.classification
and pycaret.regression
.logs.log
file cannot be created when setup
is initialized, no exception will be raised now (support for more configurable logging to come in the future)