"""
The ``mlflow.pmdarima`` module provides an API for logging and loading ``pmdarima`` models.
This module exports univariate ``pmdarima`` models in the following formats:
Pmdarima format
Serialized instance of a ``pmdarima`` model using pickle.
:py:mod:`mlflow.pyfunc`
Produced for use by generic pyfunc-based deployment tools and for batch auditing
of historical forecasts.
.. code-block:: python
:caption: Example
import pandas as pd
import mlflow
import mlflow.pyfunc
import pmdarima
from pmdarima import auto_arima
# Define a custom model class
class PmdarimaWrapper(mlflow.pyfunc.PythonModel):
def load_context(self, context):
self.model = context.artifacts["model"]
def predict(self, context, model_input):
return self.model.predict(n_periods=model_input.shape[0])
# Specify locations of source data and the model artifact
SOURCE_DATA = "https://raw.githubusercontent.com/facebook/prophet/master/examples/example_retail_sales.csv"
ARTIFACT_PATH = "model"
# Read data and recode columns
sales_data = pd.read_csv(SOURCE_DATA)
sales_data.rename(columns={"y": "sales", "ds": "date"}, inplace=True)
# Split the data into train/test
train_size = int(0.8 * len(sales_data))
train, _ = sales_data[:train_size], sales_data[train_size:]
# Create the model
model = pmdarima.auto_arima(train["sales"], seasonal=True, m=12)
# Log the model
with mlflow.start_run():
wrapper = PmdarimaWrapper()
mlflow.pyfunc.log_model(
name="model",
python_model=wrapper,
artifacts={"model": mlflow.pyfunc.model_to_dict(model)},
)
.. _Pmdarima:
http://alkaline-ml.com/pmdarima/
"""
import logging
import os
import pickle
import warnings
from typing import Any, Optional
import pandas as pd
import yaml
from packaging.version import Version
import mlflow
from mlflow import pyfunc
from mlflow.exceptions import MlflowException
from mlflow.models import Model, ModelInputExample, ModelSignature
from mlflow.models.model import MLMODEL_FILE_NAME
from mlflow.models.signature import _infer_signature_from_input_example
from mlflow.models.utils import _save_example
from mlflow.protos.databricks_pb2 import INVALID_PARAMETER_VALUE
from mlflow.tracking._model_registry import DEFAULT_AWAIT_MAX_SLEEP_SECONDS
from mlflow.tracking.artifact_utils import _download_artifact_from_uri
from mlflow.utils.docstring_utils import LOG_MODEL_PARAM_DOCS, format_docstring
from mlflow.utils.environment import (
_CONDA_ENV_FILE_NAME,
_CONSTRAINTS_FILE_NAME,
_PYTHON_ENV_FILE_NAME,
_REQUIREMENTS_FILE_NAME,
_mlflow_conda_env,
_process_conda_env,
_process_pip_requirements,
_PythonEnv,
_validate_env_arguments,
)
from mlflow.utils.file_utils import get_total_file_size, write_to
from mlflow.utils.model_utils import (
_add_code_from_conf_to_system_path,
_get_flavor_configuration,
_validate_and_copy_code_paths,
_validate_and_prepare_target_save_path,
)
from mlflow.utils.requirements_utils import _get_pinned_requirement
FLAVOR_NAME = "pmdarima"
_MODEL_BINARY_KEY = "data"
_MODEL_BINARY_FILE_NAME = "model.pmd"
_MODEL_TYPE_KEY = "model_type"
_logger = logging.getLogger(__name__)
[docs]def get_default_pip_requirements():
"""
Returns:
A list of default pip requirements for MLflow Models produced by this flavor. Calls to
:func:`save_model()` and :func:`log_model()` produce a pip environment that, at a minimum,
contains these requirements.
"""
return [_get_pinned_requirement("pmdarima")]
[docs]def get_default_conda_env():
"""
Returns:
The default Conda environment for MLflow Models produced by calls to
:func:`save_model()` and :func:`log_model()`.
"""
return _mlflow_conda_env(additional_pip_deps=get_default_pip_requirements())
[docs]@format_docstring(LOG_MODEL_PARAM_DOCS.format(package_name=FLAVOR_NAME))
def save_model(
pmdarima_model,
path,
conda_env=None,
code_paths=None,
mlflow_model=None,
signature: ModelSignature = None,
input_example: ModelInputExample = None,
pip_requirements=None,
extra_pip_requirements=None,
metadata=None,
):
"""
Save a pmdarima ``ARIMA`` model or ``Pipeline`` object to a path on the local file system.
Args:
pmdarima_model: pmdarima ``ARIMA`` or ``Pipeline`` model that has been ``fit`` on a
temporal series.
path: Local path destination for the serialized model (in pickle format) is to be saved.
conda_env: {{ conda_env }}
code_paths: {{ code_paths }}
mlflow_model: :py:mod:`mlflow.models.Model` this flavor is being added to.
signature: an instance of the :py:class:`ModelSignature <mlflow.models.ModelSignature>`
class that describes the model's inputs and outputs. If not specified but an
``input_example`` is supplied, a signature will be automatically inferred
based on the supplied input example and model. To disable automatic signature
inference when providing an input example, set ``signature`` to ``False``.
To manually infer a model signature, call
:py:func:`infer_signature() <mlflow.models.infer_signature>` on datasets
with valid model inputs, such as a training dataset with the target column
omitted, and valid model outputs, like model predictions made on the training
dataset, for example:
.. code-block:: python
from mlflow.models import infer_signature
model = pmdarima.auto_arima(data)
predictions = model.predict(n_periods=30, return_conf_int=False)
signature = infer_signature(data, predictions)
.. Warning:: if utilizing confidence interval generation in the ``predict``
method of a ``pmdarima`` model (``return_conf_int=True``), the signature
will not be inferred due to the complex tuple return type when using the
native ``ARIMA.predict()`` API. ``infer_schema`` will function correctly
if using the ``pyfunc`` flavor of the model, though.
input_example: {{ input_example }}
pip_requirements: {{ pip_requirements }}
extra_pip_requirements: {{ extra_pip_requirements }}
metadata: {{ metadata }}
.. code-block:: python
:caption: Example
import pandas as pd
import mlflow
import pmdarima
# Specify locations of source data and the model artifact
SOURCE_DATA = "https://raw.githubusercontent.com/facebook/prophet/master/examples/example_retail_sales.csv"
ARTIFACT_PATH = "model"
# Read data and recode columns
sales_data = pd.read_csv(SOURCE_DATA)
sales_data.rename(columns={"y": "sales", "ds": "date"}, inplace=True)
# Split the data into train/test
train_size = int(0.8 * len(sales_data))
train = sales_data[:train_size]
test = sales_data[train_size:]
with mlflow.start_run():
# Create the model
model = pmdarima.auto_arima(train["sales"], seasonal=True, m=12)
# Save the model to the specified path
mlflow.pmdarima.save_model(model, "model")
"""
import pmdarima
_validate_env_arguments(conda_env, pip_requirements, extra_pip_requirements)
path = os.path.abspath(path)
_validate_and_prepare_target_save_path(path)
code_dir_subpath = _validate_and_copy_code_paths(code_paths, path)
if mlflow_model is None:
mlflow_model = Model()
saved_example = _save_example(mlflow_model, input_example, path)
if signature is None and saved_example is not None:
wrapped_model = _PmdarimaModelWrapper(pmdarima_model)
signature = _infer_signature_from_input_example(saved_example, wrapped_model)
elif signature is False:
signature = None
if signature is not None:
mlflow_model.signature = signature
if metadata is not None:
mlflow_model.metadata = metadata
model_data_path = os.path.join(path, _MODEL_BINARY_FILE_NAME)
_save_model(pmdarima_model, model_data_path)
model_bin_kwargs = {_MODEL_BINARY_KEY: _MODEL_BINARY_FILE_NAME}
pyfunc.add_to_model(
mlflow_model,
loader_module="mlflow.pmdarima",
conda_env=_CONDA_ENV_FILE_NAME,
python_env=_PYTHON_ENV_FILE_NAME,
code=code_dir_subpath,
**model_bin_kwargs,
)
flavor_conf = {
_MODEL_TYPE_KEY: pmdarima_model.__class__.__name__,
**model_bin_kwargs,
}
mlflow_model.add_flavor(
FLAVOR_NAME, pmdarima_version=pmdarima.__version__, code=code_dir_subpath, **flavor_conf
)
if size := get_total_file_size(path):
mlflow_model.model_size_bytes = size
mlflow_model.save(os.path.join(path, MLMODEL_FILE_NAME))
if conda_env is None:
if pip_requirements is None:
default_reqs = get_default_pip_requirements()
inferred_reqs = mlflow.models.infer_pip_requirements(
path, FLAVOR_NAME, fallback=default_reqs
)
default_reqs = sorted(set(inferred_reqs).union(default_reqs))
else:
default_reqs = None
conda_env, pip_requirements, pip_constraints = _process_pip_requirements(
default_reqs, pip_requirements, extra_pip_requirements
)
else:
conda_env, pip_requirements, pip_constraints = _process_conda_env(conda_env)
with open(os.path.join(path, _CONDA_ENV_FILE_NAME), "w") as f:
yaml.safe_dump(conda_env, stream=f, default_flow_style=False)
if pip_constraints:
write_to(os.path.join(path, _CONSTRAINTS_FILE_NAME), "\n".join(pip_constraints))
write_to(os.path.join(path, _REQUIREMENTS_FILE_NAME), "\n".join(pip_requirements))
_PythonEnv.current().to_yaml(os.path.join(path, _PYTHON_ENV_FILE_NAME))
[docs]@format_docstring(LOG_MODEL_PARAM_DOCS.format(package_name=FLAVOR_NAME))
def log_model(
pmdarima_model,
artifact_path: Optional[str] = None,
conda_env=None,
code_paths=None,
registered_model_name=None,
signature: ModelSignature = None,
input_example: ModelInputExample = None,
await_registration_for=DEFAULT_AWAIT_MAX_SLEEP_SECONDS,
pip_requirements=None,
extra_pip_requirements=None,
metadata=None,
name: Optional[str] = None,
params: Optional[dict[str, Any]] = None,
tags: Optional[dict[str, Any]] = None,
model_type: Optional[str] = None,
step: int = 0,
model_id: Optional[str] = None,
**kwargs,
):
"""
Logs a ``pmdarima`` ``ARIMA`` or ``Pipeline`` object as an MLflow artifact for the current run.
Args:
pmdarima_model: pmdarima ``ARIMA`` or ``Pipeline`` model that has been ``fit`` on a
temporal series.
artifact_path: Deprecated. Use `name` instead.
conda_env: {{ conda_env }}
code_paths: {{ code_paths }}
registered_model_name: If given, create a model
version under ``registered_model_name``, also creating a
registered model if one with the given name does not exist.
signature: an instance of the :py:class:`ModelSignature <mlflow.models.ModelSignature>`
class that describes the model's inputs and outputs. If not specified but an
``input_example`` is supplied, a signature will be automatically inferred
based on the supplied input example and model. To disable automatic signature
inference when providing an input example, set ``signature`` to ``False``.
To manually infer a model signature, call
:py:func:`infer_signature() <mlflow.models.infer_signature>` on datasets
with valid model inputs, such as a training dataset with the target column
omitted, and valid model outputs, like model predictions made on the training
dataset, for example:
.. code-block:: python
from mlflow.models import infer_signature
model = pmdarima.auto_arima(data)
predictions = model.predict(n_periods=30, return_conf_int=False)
signature = infer_signature(data, predictions)
.. Warning:: if utilizing confidence interval generation in the ``predict``
method of a ``pmdarima`` model (``return_conf_int=True``), the signature
will not be inferred due to the complex tuple return type when using the
native ``ARIMA.predict()`` API. ``infer_schema`` will function correctly
if using the ``pyfunc`` flavor of the model, though.
input_example: {{ input_example }}
await_registration_for: Number of seconds to wait for the model version
to finish being created and is in ``READY`` status.
By default, the function waits for five minutes.
Specify 0 or None to skip waiting.
pip_requirements: {{ pip_requirements }}
extra_pip_requirements: {{ extra_pip_requirements }}
metadata: {{ metadata }}
name: {{ name }}
params: {{ params }}
tags: {{ tags }}
model_type: {{ model_type }}
step: {{ step }}
model_id: {{ model_id }}
kwargs: Additional arguments for :py:class:`mlflow.models.model.Model`
Returns:
A :py:class:`ModelInfo <mlflow.models.model.ModelInfo>` instance that contains the
metadata of the logged model.
.. code-block:: python
:caption: Example
import pandas as pd
import mlflow
from mlflow.models import infer_signature
import pmdarima
from pmdarima.metrics import smape
# Specify locations of source data and the model artifact
SOURCE_DATA = "https://raw.githubusercontent.com/facebook/prophet/master/examples/example_retail_sales.csv"
ARTIFACT_PATH = "model"
# Read data and recode columns
sales_data = pd.read_csv(SOURCE_DATA)
sales_data.rename(columns={"y": "sales", "ds": "date"}, inplace=True)
# Split the data into train/test
train_size = int(0.8 * len(sales_data))
train = sales_data[:train_size]
test = sales_data[train_size:]
with mlflow.start_run():
# Create the model
model = pmdarima.auto_arima(train["sales"], seasonal=True, m=12)
# Calculate metrics
prediction = model.predict(n_periods=len(test))
metrics = {"smape": smape(test["sales"], prediction)}
# Infer signature
input_sample = pd.DataFrame(train["sales"])
output_sample = pd.DataFrame(model.predict(n_periods=5))
signature = infer_signature(input_sample, output_sample)
# Log model
mlflow.pmdarima.log_model(model, name=ARTIFACT_PATH, signature=signature)
"""
return Model.log(
artifact_path=artifact_path,
name=name,
flavor=mlflow.pmdarima,
registered_model_name=registered_model_name,
pmdarima_model=pmdarima_model,
conda_env=conda_env,
code_paths=code_paths,
signature=signature,
input_example=input_example,
await_registration_for=await_registration_for,
pip_requirements=pip_requirements,
extra_pip_requirements=extra_pip_requirements,
metadata=metadata,
params=params,
tags=tags,
model_type=model_type,
step=step,
model_id=model_id,
**kwargs,
)
[docs]def load_model(model_uri, dst_path=None):
"""
Load a ``pmdarima`` ``ARIMA`` model or ``Pipeline`` object from a local file or a run.
Args:
model_uri: The location, in URI format, of the MLflow model. For example:
- ``/Users/me/path/to/local/model``
- ``relative/path/to/local/model``
- ``s3://my_bucket/path/to/model``
- ``runs:/<mlflow_run_id>/run-relative/path/to/model``
- ``mlflow-artifacts:/path/to/model``
For more information about supported URI schemes, see
`Referencing Artifacts <https://www.mlflow.org/docs/latest/tracking.html#
artifact-locations>`_.
dst_path: The local filesystem path to which to download the model artifact.
This directory must already exist. If unspecified, a local output
path will be created.
Returns:
A ``pmdarima`` model instance
.. code-block:: python
:caption: Example
import pandas as pd
import mlflow
from mlflow.models import infer_signature
import pmdarima
from pmdarima.metrics import smape
# Specify locations of source data and the model artifact
SOURCE_DATA = "https://raw.githubusercontent.com/facebook/prophet/master/examples/example_retail_sales.csv"
ARTIFACT_PATH = "model"
# Read data and recode columns
sales_data = pd.read_csv(SOURCE_DATA)
sales_data.rename(columns={"y": "sales", "ds": "date"}, inplace=True)
# Split the data into train/test
train_size = int(0.8 * len(sales_data))
train = sales_data[:train_size]
test = sales_data[train_size:]
with mlflow.start_run():
# Create the model
model = pmdarima.auto_arima(train["sales"], seasonal=True, m=12)
# Calculate metrics
prediction = model.predict(n_periods=len(test))
metrics = {"smape": smape(test["sales"], prediction)}
# Infer signature
input_sample = pd.DataFrame(train["sales"])
output_sample = pd.DataFrame(model.predict(n_periods=5))
signature = infer_signature(input_sample, output_sample)
# Log model
input_example = input_sample.head()
mlflow.pmdarima.log_model(
model, name=ARTIFACT_PATH, signature=signature, input_example=input_example
)
# Get the model URI for loading
model_uri = mlflow.get_artifact_uri(ARTIFACT_PATH)
# Load the model
loaded_model = mlflow.pmdarima.load_model(model_uri)
# Forecast for the next 60 days
forecast = loaded_model.predict(n_periods=60)
print(f"forecast: {forecast}")
.. code-block:: text
:caption: Output
forecast:
234 382452.397246
235 380639.458720
236 359805.611219
...
"""
local_model_path = _download_artifact_from_uri(artifact_uri=model_uri, output_path=dst_path)
flavor_conf = _get_flavor_configuration(model_path=local_model_path, flavor_name=FLAVOR_NAME)
_add_code_from_conf_to_system_path(local_model_path, flavor_conf)
pmdarima_model_file_path = os.path.join(
local_model_path, flavor_conf.get(_MODEL_BINARY_KEY, _MODEL_BINARY_FILE_NAME)
)
return _load_model(pmdarima_model_file_path)
def _save_model(model, path):
with open(path, "wb") as f:
pickle.dump(model, f)
def _load_model(path):
with open(path, "rb") as pickled_model:
return pickle.load(pickled_model)
def _load_pyfunc(path):
return _PmdarimaModelWrapper(_load_model(path))
class _PmdarimaModelWrapper:
def __init__(self, pmdarima_model):
import pmdarima
self.pmdarima_model = pmdarima_model
self._pmdarima_version = pmdarima.__version__
def get_raw_model(self):
"""
Returns the underlying model.
"""
return self.pmdarima_model
def predict(self, dataframe, params: Optional[dict[str, Any]] = None) -> pd.DataFrame:
"""
Args:
dataframe: Model input data.
params: Additional parameters to pass to the model for inference.
Returns:
Model predictions.
"""
df_schema = dataframe.columns.values.tolist()
if len(dataframe) > 1:
raise MlflowException(
f"The provided prediction pd.DataFrame contains {len(dataframe)} rows. "
"Only 1 row should be supplied.",
error_code=INVALID_PARAMETER_VALUE,
)
attrs = dataframe.to_dict(orient="index").get(0)
n_periods = attrs.get("n_periods", None)
if not n_periods:
raise MlflowException(
f"The provided prediction configuration pd.DataFrame columns ({df_schema}) do not "
"contain the required column `n_periods` for specifying future prediction periods "
"to generate.",
error_code=INVALID_PARAMETER_VALUE,
)
if not isinstance(n_periods, int):
raise MlflowException(
f"The provided `n_periods` value {n_periods} must be an integer."
f"provided type: {type(n_periods)}",
error_code=INVALID_PARAMETER_VALUE,
)
# NB Any model that is trained with exogenous regressor elements will need to provide
# `X` entries as a 2D array structure to the predict method.
exogenous_regressor = attrs.get("X", None)
if exogenous_regressor and Version(self._pmdarima_version) < Version("1.8.0"):
warnings.warn(
"An exogenous regressor element was provided in column 'X'. This is "
"supported only in pmdarima version >= 1.8.0. Installed version: "
f"{self._pmdarima_version}"
)
return_conf_int = attrs.get("return_conf_int", False)
alpha = attrs.get("alpha", 0.05)
if not isinstance(n_periods, int):
raise MlflowException(
"The prediction DataFrame must contain a column `n_periods` with "
"an integer value for number of future periods to predict.",
error_code=INVALID_PARAMETER_VALUE,
)
if Version(self._pmdarima_version) >= Version("1.8.0"):
raw_predictions = self.pmdarima_model.predict(
n_periods=n_periods,
X=exogenous_regressor,
return_conf_int=return_conf_int,
alpha=alpha,
)
else:
raw_predictions = self.pmdarima_model.predict(
n_periods=n_periods,
return_conf_int=return_conf_int,
alpha=alpha,
)
if return_conf_int:
ci_low, ci_high = list(zip(*raw_predictions[1]))
predictions = pd.DataFrame.from_dict(
{"yhat": raw_predictions[0], "yhat_lower": ci_low, "yhat_upper": ci_high}
)
else:
predictions = pd.DataFrame.from_dict({"yhat": raw_predictions})
return predictions