Snowflake-ml-python

New Features

- Model Registry: Added support save/load/deploy xgboost Booster model.
- Model Registry: Added support to get the model name and the model version from model references.

Bug Fixes

- Model Registry: Restore the db/schema back to the session after `create_model_registry()`.
- Model Registry: Fixed an issue that the UDF name created when deploying a model is not identical to what is provided
and cannot be correctly dropped when deployment getting dropped.
- connection_params.SnowflakeLoginOptions(): Added support for `private_key_path`.

New Features

- Model Registry: Added support save/load/deploy Tensorflow models (`tensorflow.Module`).
- Model Registry: Added support save/load/deploy MLFlow PyFunc models (`mlflow.pyfunc.PyFuncModel`).
- Model Development: Input dataframes can now be joined against data loaded from staged files.
- Model Development: Added support for non-English languages.

Bug Fixes

- Model Registry: Fix an issue that model dependencies are incorrectly reported as unresolvable on certain platforms.

Behavior Changes

- Model Registry: When predicting a model whose output is a list of NumPy ndarray, the output would not be flattened,
instead, every ndarray will act as a feature(column) in the output.

New Features

- Model Registry: Added support save/load/deploy PyTorch models (`torch.nn.Module` and `torch.jit.ScriptModule`).

Bug Fixes

- Model Registry: Fix an issue that when database or schema name provided to `create_model_registry` contains special
characters, the model registry cannot be created.
- Model Registry: Fix an issue that `get_model_description` returns with additional quotes.
- Model Registry: Fix incorrect error message when attempting to remove a unset tag of a model.
- Model Registry: Fix a typo in the default deployment table name.
- Model Registry: Snowpark dataframe for sample input or input for `predict` method that contains a column with
Snowflake `NUMBER(precision, scale)` data type where `scale = 0` will not lead to error, and will now correctly
recognized as `INT64` data type in model signature.
- Model Registry: Fix an issue that prevent model logged in the system whose default encoding is not UTF-8 compatible
from deploying.
- Model Registry: Added earlier and better error message when any file name in the model or the file name of model
itself contains characters that are unable to be encoded using ASCII. It is currently not supported to deploy such a
model.

Behavior Changes

- Model Registry: Prohibit non-snowflake-native models from being logged.
- Model Registry: `_use_local_snowml` parameter in options of `deploy()` has been removed.
- Model Registry: A default `False` `embed_local_ml_library` parameter has been added to the options of `log_model()`.
With this set to `False` (default), the version of the local snowflake-ml-python library will be recorded and used when
deploying the model. With this set to `True`, local snowflake-ml-python library will be embedded into the logged model,
and will be used when you load or deploy the model.

New Features

- Model Registry: A new optional argument named `code_paths` has been added to the arguments of `log_model()` for users
to specify additional code paths to be imported when loading and deploying the model.
- Model Registry: A new optional argument named `options` has been added to the arguments of `log_model()` to specify
any additional options when saving the model.
- Model Development: Added metrics:
- d2_absolute_error_score
- d2_pinball_score
- explained_variance_score
- mean_absolute_error
- mean_absolute_percentage_error
- mean_squared_error

Bug Fixes

- Model Development: `accuracy_score()` now works when given label column names are lists of a single value.

Behavior Changes

- Model Development: Changed Metrics APIs to imitate sklearn metrics modules:
- `accuracy_score()`, `confusion_matrix()`, `precision_recall_fscore_support()`, `precision_score()` methods move from
respective modules to `metrics.classification`.
- Model Registry: The default table/stage created by the Registry now uses "_SYSTEM_" as a prefix.
- Model Registry: `get_model_history()` method as been enhanced to include the history of model deployment.

New Features

- Model Registry: A default `False` flag named `replace_udf` has been added to the options of `deploy()`. Setting this
to `True` will allow overwrite existing UDF with the same name when deploying.
- Model Development: Added metrics:
- f1_score
- fbeta_score
- recall_score
- roc_auc_score
- roc_curve
- log_loss
- precision_recall_curve
- Model Registry: A new argument named `permanent` has been added to the argument of `deploy()`. Setting this to `True`
allows the creation of a permanent deployment without needing to specify the UDF location.
- Model Registry: A new method `list_deployments()` has been added to enumerate all permanent deployments originating
from a specific model.
- Model Registry: A new method `get_deployment()` has been added to fetch a deployment by its deployment name.
- Model Registry: A new method `delete_deployment()` has been added to remove an existing permanent deployment.

Behavior Changes

- Model Registry: `predict()` method moves from Registry to ModelReference.
- Model Registry: `_snowml_wheel_path` parameter in options of `deploy()`, is replaced with `_use_local_snowml` with
default value of `False`. Setting this to `True` will have the same effect of uploading local SnowML code when executing
model in the warehouse.
- Model Registry: Removed `id` field from `ModelReference` constructor.
- Model Development: Preprocessing and Metrics move to the modeling package: `snowflake.ml.modeling.preprocessing` and
`snowflake.ml.modeling.metrics`.
- Model Development: `get_sklearn_object()` method is renamed to `to_sklearn()`, `to_xgboost()`, and `to_lightgbm()` for
respective native models.

New Features

- Added PolynomialFeatures transformer to the snowflake.ml.modeling.preprocessing module.
- Added metrics:
- accuracy_score
- confusion_matrix
- precision_recall_fscore_support
- precision_score

Bug Fixes

- Model Registry: Model version can now be any string (not required to be a valid identifier)
- Model Deployment: `deploy()` & `predict()` methods now correctly escapes identifiers