probinet.evaluation.link_prediction

Functions for evaluating link prediction.

Functions

calculate_f1_score(data0, pred[, mask, ...])	Calculate the F1 score for the given predictions and data.
compute_AUC_from_ranked_predictions(...)	Calculate the AUC (Area Under the Curve) for the given ranked list of predictions.
compute_L1loss(X, Xtilde)	Calculate the L1 loss between two matrices.
compute_link_prediction_AUC(data0, pred[, mask])	Calculate the AUC (Area Under the Curve) for link prediction.
compute_multilayer_link_prediction_AUC(B, u, ...)	Calculate the AUC (Area Under the Curve) for link prediction in multilayer data.
mask_or_flatten_array(-> ~numpy.ndarray)

probinet.evaluation.link_prediction.calculate_f1_score(data0: ndarray, pred: ndarray, mask: ndarray | None = None, threshold: float = 0.1) → float

Calculate the F1 score for the given predictions and data.

Parameters:

• data0 (np.ndarray) – The original adjacency matrix.

• pred (np.ndarray) – The predicted adjacency matrix.

• mask (Optional[np.ndarray], optional) – The mask to apply on the data, by default None.

• threshold (float, optional) – The threshold to binarize the predictions, by default 0.1.

Returns:

The F1 score for the given predictions and data.

Return type:

float

probinet.evaluation.link_prediction.compute_AUC_from_ranked_predictions(ranked_predictions: list[tuple[float, int]], num_positive_samples: int, num_negative_samples: int) → float

Calculate the AUC (Area Under the Curve) for the given ranked list of predictions.

Parameters:

• ranked_predictions (list[tuple[float, int]]) – The ranked list of predictions, where each tuple contains a score and the actual value.

• num_positive_samples (int) – The number of positive samples.

• num_negative_samples (int) – The number of negative samples.

Returns:

The AUC value for the ranked predictions.

Return type:

float

probinet.evaluation.link_prediction.compute_L1loss(X: ndarray, Xtilde: ndarray) → float

Calculate the L1 loss between two matrices.

Parameters:

• X (np.ndarray) – The first matrix.

• Xtilde (np.ndarray) – The second matrix to compare against the first matrix.

Returns:

The L1 loss between the two matrices, rounded to three decimal places.

Return type:

float

probinet.evaluation.link_prediction.compute_link_prediction_AUC(data0: ndarray, pred: ndarray, mask: ndarray | None = None) → float

Calculate the AUC (Area Under the Curve) for link prediction.

Parameters:

• data0 (np.ndarray) – The original adjacency matrix.

• pred (np.ndarray) – The predicted adjacency matrix.

• mask (Optional[np.ndarray], optional) – The mask to apply on the data, by default None.

Returns:

The AUC value for the link prediction.

Return type:

float

probinet.evaluation.link_prediction.compute_multilayer_link_prediction_AUC(B: ndarray, u: ndarray, v: ndarray, w: ndarray, mask: ndarray | None = None) → float

Calculate the AUC (Area Under the Curve) for link prediction in multilayer data.

Parameters:

• B (np.ndarray) – The original adjacency tensor.

• u (np.ndarray) – The first factor matrix.

• v (np.ndarray) – The second factor matrix.

• w (np.ndarray) – The third factor matrix.

• mask (Optional[np.ndarray], optional) – The mask to apply on the data, by default None.

Returns:

The AUC value for the link prediction in multilayer data.

Return type:

float

probinet.evaluation.link_prediction.mask_or_flatten_array(mask: ndarray | None, expected_adjacency: ndarray) → ndarray

probinet.evaluation.link_prediction.mask_or_flatten_array(mask: None, expected_adjacency: ndarray) → ndarray

probinet.evaluation.link_prediction.mask_or_flatten_array(mask: ndarray, expected_adjacency: ndarray) → ndarray