# Author: Niels Nuyttens <niels@nannyml.com>
#
# License: Apache Software License 2.0
import abc
import logging
from logging import Logger
from typing import Callable, Dict, List, Optional, Tuple, Type, Union
import numpy as np
import pandas as pd
from nannyml._typing import ProblemType
from nannyml.chunk import Chunker
from nannyml.exceptions import InvalidArgumentsException
from nannyml.thresholds import Threshold, calculate_threshold_values
[docs]class Metric(abc.ABC):
"""A performance metric used to calculate realized model performance."""
def __init__(
self,
name: str,
y_true: str,
y_pred: str,
components: List[Tuple[str, str]],
threshold: Threshold,
y_pred_proba: Optional[Union[str, Dict[str, str]]] = None,
upper_threshold_limit: Optional[float] = None,
lower_threshold_limit: Optional[float] = None,
**kwargs,
):
"""Creates a new Metric instance.
Parameters
----------
components: List[Tuple[str, str]]
A list of (display_name, column_name) tuples. The
display_name is used for display purposes, while the
column_name is used for column names in the output.
upper_threshold_limit : float, default=None
An optional upper threshold for the performance metric.
lower_threshold_limit : float, default=None
An optional lower threshold for the performance metric.
"""
self.name: str = name
self.y_true = y_true
self.y_pred = y_pred
self.y_pred_proba = y_pred_proba
self.threshold = threshold
self.upper_threshold_value: Optional[float] = None
self.lower_threshold_value: Optional[float] = None
self.lower_threshold_value_limit: Optional[float] = lower_threshold_limit
self.upper_threshold_value_limit: Optional[float] = upper_threshold_limit
# A list of (display_name, column_name) tuples
self.components: List[Tuple[str, str]] = components
@property
def _logger(self) -> logging.Logger:
return logging.getLogger(__name__)
[docs] def fit(self, reference_data: pd.DataFrame, chunker: Chunker):
"""Fits a Metric on reference data.
Parameters
----------
reference_data: pd.DataFrame
The reference data used for fitting. Must have target data available.
chunker: Chunker
The :class:`~nannyml.chunk.Chunker` used to split the reference data into chunks.
This value is provided by the calling
:class:`~nannyml.performance_calculation.calculator.PerformanceCalculator`.
"""
self._fit(reference_data)
# Calculate alert thresholds
reference_chunk_results = np.asarray([self.calculate(chunk.data) for chunk in chunker.split(reference_data)])
self.lower_threshold_value, self.upper_threshold_value = calculate_threshold_values(
threshold=self.threshold,
data=reference_chunk_results,
lower_threshold_value_limit=self.lower_threshold_value_limit,
upper_threshold_value_limit=self.upper_threshold_value_limit,
logger=self._logger,
metric_name=self.display_name,
)
return
def _fit(self, reference_data: pd.DataFrame):
raise NotImplementedError(
f"'{self.__class__.__name__}' is a subclass of Metric and it must implement the _fit method"
)
[docs] def calculate(self, data: pd.DataFrame):
"""Calculates performance metrics on data.
Parameters
----------
data: pd.DataFrame
The data to calculate performance metrics on. Requires presence of either the predicted labels or
prediction scores/probabilities (depending on the metric to be calculated), as well as the target data.
"""
return self._calculate(data)
def _calculate(self, data: pd.DataFrame):
raise NotImplementedError(
f"'{self.__class__.__name__}' is a subclass of Metric and it must implement the _calculate method"
)
[docs] def sampling_error(self, data: pd.DataFrame):
"""Calculates the sampling error with respect to the reference data for a given chunk of data.
Parameters
----------
data: pd.DataFrame
The data to calculate the sampling error on, with respect to the reference data.
Returns
-------
sampling_error: float
The expected sampling error.
"""
return self._sampling_error(data)
def _sampling_error(self, data: pd.DataFrame):
raise NotImplementedError(
f"'{self.__class__.__name__}' is a subclass of Metric and it must implement the _sampling_error method"
)
[docs] def alert(self, value: float) -> bool:
return (self.lower_threshold_value is not None and value < self.lower_threshold_value) or (
self.upper_threshold_value is not None and value > self.upper_threshold_value
)
[docs] def __eq__(self, other):
"""Establishes equality by comparing all properties."""
return (
self.display_name == other.display_name
and self.column_name == other.column_name
and self.components == other.components
and self.upper_threshold_value == other.upper_threshold_value
and self.lower_threshold_value == other.lower_threshold_value
)
[docs] def get_chunk_record(self, chunk_data: pd.DataFrame) -> Dict:
"""Returns a DataFrame containing the performance metrics for a given chunk."""
if len(self.components) > 1:
raise NotImplementedError(
"cannot use default 'get_chunk_record' implementation when a metric has multiple components."
)
column_name = self.components[0][1]
chunk_record = {}
realized_value = self.calculate(chunk_data)
sampling_error = self.sampling_error(chunk_data)
chunk_record[f'{column_name}_sampling_error'] = sampling_error
chunk_record[f'{column_name}'] = realized_value
chunk_record[f'{column_name}_upper_threshold'] = self.upper_threshold_value
chunk_record[f'{column_name}_lower_threshold'] = self.lower_threshold_value
chunk_record[f'{column_name}_alert'] = self.alert(realized_value)
return chunk_record
@property
def display_name(self) -> str:
return self.name
@property
def column_name(self) -> str:
return self.components[0][1]
@property
def display_names(self) -> List[str]:
return [c[0] for c in self.components]
@property
def column_names(self) -> List[str]:
return [c[1] for c in self.components]
[docs]class MetricFactory:
"""A factory class that produces Metric instances based on a given magic string or a metric specification."""
registry: Dict[str, Dict[ProblemType, Type[Metric]]] = {}
@classmethod
def _logger(cls) -> Logger:
return logging.getLogger(__name__)
[docs] @classmethod
def create(cls, key: str, use_case: ProblemType, **kwargs) -> Metric:
"""Returns a Metric instance for a given key."""
if not isinstance(key, str):
raise InvalidArgumentsException(
f"cannot create metric given a '{type(key)}'" "Please provide a string, function or Metric"
)
if key not in cls.registry:
raise InvalidArgumentsException(
f"unknown metric key '{key}' given. "
"Should be one of ['roc_auc', 'f1', 'precision', 'recall', 'specificity', "
"'accuracy', 'confusion_matrix', 'business_value']."
)
if use_case not in cls.registry[key]:
raise RuntimeError(
f"metric '{key}' is currently not supported for use case {use_case}. "
"Please specify another metric or use one of these supported model types for this metric: "
f"{[md.value for md in cls.registry[key]]}"
)
metric_class = cls.registry[key][use_case]
return metric_class(**kwargs)
[docs] @classmethod
def register(cls, metric: str, use_case: ProblemType) -> Callable:
def inner_wrapper(wrapped_class: Type[Metric]) -> Type[Metric]:
if metric in cls.registry:
if use_case in cls.registry[metric]:
cls._logger().warning(f"re-registering Metric for metric='{metric}' and use_case='{use_case}'")
cls.registry[metric][use_case] = wrapped_class
else:
cls.registry[metric] = {use_case: wrapped_class}
return wrapped_class
return inner_wrapper
def _common_data_cleaning(y_true, y_pred):
y_true, y_pred = (
pd.Series(y_true).reset_index(drop=True),
pd.Series(y_pred).reset_index(drop=True),
)
y_true = y_true[~y_pred.isna()]
y_pred.dropna(inplace=True)
y_pred = y_pred[~y_true.isna()]
y_true.dropna(inplace=True)
return y_true, y_pred