Source code for denspp.offline.metric.data_torch

from sklearn.metrics import precision_recall_fscore_support
from torch import Tensor, tensor, sum, eq, ne




def calculate_number_true_predictions(pred: Tensor, true: Tensor) -> Tensor:
    """Function for determining the true predicted values
    Args:
        pred:   Tensor with predicted values from model
        true:   Tensor with true labels from dataset
    Return
        Tensor with metric
    """
    assert pred.shape == true.shape, "Dimension / shape mismatch"
    return sum(eq(pred, true))





def calculate_number_false_predictions(pred: Tensor, true: Tensor) -> Tensor:
    """Function for determining the true predicted values
    Args:
        pred:   Tensor with predicted values from model
        true:   Tensor with true labels from dataset
    Return
        Tensor with metric
    """
    assert pred.shape == true.shape, "Dimension / shape mismatch"
    return sum(ne(pred, true))





def calculate_precision(pred: Tensor, true: Tensor) -> Tensor:
    """Function for determining the precision metric
    Args:
        pred:   Tensor with predicted values from model
        true:   Tensor with true labels from dataset
    Return
        Tensor with metrics [precision]
    """
    assert pred.shape == true.shape, "Dimension / shape mismatch"
    return tensor(precision_recall_fscore_support(true, pred, average="micro", warn_for=tuple())[0])





def calculate_recall(pred: Tensor, true: Tensor) -> Tensor:
    """Function for determining the precision metric
    Args:
        pred:   Tensor with predicted values from model
        true:   Tensor with true labels from dataset
    Return
        Tensor with metrics [precision]
    """
    assert pred.shape == true.shape, "Dimension / shape mismatch"
    return tensor(precision_recall_fscore_support(true, pred, average="micro", warn_for=tuple())[1])





def calculate_fbeta(pred: Tensor, true: Tensor, beta: float=1.0) -> Tensor:
    """Function for determining the precision metric
    Args:
        pred:   Tensor with predicted values from model
        true:   Tensor with true labels from dataset
        beta:   Beta value for getting Fbeta metric
    Return
        Tensor with metrics [precision]
    """
    assert pred.shape == true.shape, "Dimension / shape mismatch"
    return tensor(precision_recall_fscore_support(true, pred, beta=beta, average="micro", warn_for=tuple())[2])