sklearn_quantile

Source code: sensai/sklearn_quantile.py

---

class RandomForestQuantileRegressorVectorRegressionModel(confidence: float, random_state=42, **kwargs)

Bases: AbstractSkLearnMultipleOneDimVectorRegressionModel

Parameters:

• q – the default quantile that is used for predictions

• kwargs – keyword arguments to pass on to RandomForestQuantileRegressor

predict_confidence_intervals(x: DataFrame, var_name: Optional[str] = None)

Parameters:

• x – the input data

• var_name – the predicted variable name; may be None if there is only one predicted variable

Returns:

an array of shape [2, N], where the first dimension contains the confidence interval’s lower bounds and the second its upper bounds

class QuantileRegressionMetric(name: Optional[str] = None, bounds: Optional[Tuple[float, float]] = None)

Bases: RegressionMetric, ABC

Parameters:

• name – the name of the metric; if None use the class’ name attribute

• bounds – the minimum and maximum values the metric can take on (or None if the bounds are not specified)

static compute_confidence_intervals(model: VectorRegressionModel, io_data: InputOutputData = None) → ndarray

name: str

class QuantileRegressionMetricAccuracyInConfidenceInterval(name: Optional[str] = None, bounds: Optional[Tuple[float, float]] = None)

Bases: QuantileRegressionMetric

Metric reflecting the accuracy of the confidence interval, i.e. the relative frequency of predictions where the confidence interval contains the ground true value

Parameters:

• name – the name of the metric; if None use the class’ name attribute

• bounds – the minimum and maximum values the metric can take on (or None if the bounds are not specified)

name: str = 'AccuracyInCI'

classmethod compute_value(y_true: ndarray, y_predicted: ndarray, model: Optional[VectorRegressionModel] = None, io_data: Optional[InputOutputData] = None)

class QuantileRegressionMetricConfidenceIntervalMeanSize(name: Optional[str] = None, bounds: Optional[Tuple[float, float]] = None)

Bases: QuantileRegressionMetric

Metric for the mean size of the confidence interval

Parameters:

• name – the name of the metric; if None use the class’ name attribute

• bounds – the minimum and maximum values the metric can take on (or None if the bounds are not specified)

name: str = 'MeanSizeCI'

classmethod compute_value(y_true: ndarray, y_predicted: ndarray, model: Optional[VectorRegressionModel] = None, io_data: Optional[InputOutputData] = None)

class QuantileRegressionMetricConfidenceIntervalMedianSize(name: Optional[str] = None, bounds: Optional[Tuple[float, float]] = None)

Bases: QuantileRegressionMetric

Metric for the median size of the confidence interval

Parameters:

• name – the name of the metric; if None use the class’ name attribute

• bounds – the minimum and maximum values the metric can take on (or None if the bounds are not specified)

name: str = 'MedianSizeCI'

classmethod compute_value(y_true: ndarray, y_predicted: ndarray, model: Optional[VectorRegressionModel] = None, io_data: Optional[InputOutputData] = None)

class QuantileRegressionMetricRelFreqMaxSizeConfidenceInterval(max_size: float)

Bases: QuantileRegressionMetric

Relative frequency of confidence interval having the given maximum size

Parameters:

• name – the name of the metric; if None use the class’ name attribute

• bounds – the minimum and maximum values the metric can take on (or None if the bounds are not specified)

compute_value(y_true: ndarray, y_predicted: ndarray, model: Optional[VectorRegressionModel] = None, io_data: Optional[InputOutputData] = None)

name: str