Add Accuracy in time

hazardous/metrics/_accuracy_in_time.py

@@ -0,0 +1,54 @@
+import numpy as np
+
+from hazardous.utils import check_y_survival
+
+
+def accuracy_in_time(y_test, y_pred, times, quantiles=None, taus=None):
+    event_true, _ = check_y_survival(y_test)
+
+    if y_pred.ndim != 3:
+        raise ValueError(
+            "'y_pred' must be a 3D array with shape (n_samples, n_events, n_times), got"
+            f" shape {y_pred.shape}."
+        )
+    if y_pred.shape[0] != event_true.shape[0]:
+        raise ValueError(
+            "'y_true' and 'y_pred' must have the same number of samples, "
+            f"got {event_true.shape[0]} and {y_pred.shape[0]} respectively."
+        )
+    times = np.atleast_1d(times)
+    if y_pred.shape[1] != times.shape[0]:
+        raise ValueError(
+            f"'times' length ({times.shape[0]}) "
+            f"must be equal to y_pred.shape[1] ({y_pred.shape[1]})."
+        )
+
+    if quantiles is not None:
+        if taus is not None:
+            raise ValueError("'quantiles' and 'taus' can't be set at the same time.")
+
+        quantiles = np.atleast_1d(quantiles)
+        if any(quantiles < 0) or any(quantiles > 1):
+            raise ValueError(f"quantiles must be in [0, 1], got {quantiles}.")
+        taus = np.quantile(times, quantiles)
+
+    elif quantiles is None and taus is None:
+        n_quantiles = min(times.shape[0], 8)
+        quantiles = np.linspace(1 / n_quantiles, 1, n_quantiles)
+        taus = np.quantile(times, quantiles)
+
+    acc_in_time = []
+
+    for tau in taus:
+        mask_past_censored = (y_test["event"] == 0) & (y_test["duration"] < tau)
+
+        tau_idx = np.searchsorted(times, tau)
+        y_pred_at_t = y_pred[:, :, tau_idx]
+        y_pred_class = y_pred_at_t[~mask_past_censored, :].argmax(axis=1)
+
+        y_test_class = y_test["event"] * (y_test["duration"] < tau)
+        y_test_class = y_test_class.loc[~mask_past_censored]
+
+        acc_in_time.append((y_test_class.values == y_pred_class).mean())
+
+    return acc_in_time, taus

hazardous/metrics/_brier_score.py

@@ -145,6 +145,7 @@ def brier_score_incidence(self, y_true, y_pred, times):
                 "'y_true' and 'y_pred' must have the same number of samples, "
                 f"got {event_true.shape[0]} and {y_pred.shape[0]} respectively."
             )
+        times = np.atleast_1d(times)
         if y_pred.shape[1] != times.shape[0]:
             raise ValueError(
                 f"'times' length ({times.shape[0]}) "