replace type aliases that are incompatible with python 3.7, make valu…

…e of 'n_init' KMeans parameter dependent on sklearn version

pypsf/clustering.py

@@ -1,8 +1,10 @@
 import numpy as np
+import sklearn
 from sklearn.cluster import KMeans
+from typing import List
 
 
-def run_clustering(cycles: list[np.array], n_clusters: int) -> KMeans:
+def run_clustering(cycles: List[np.array] or np.array, n_clusters: int) -> KMeans:
     """
     Apply K-means clustering to the provided list of cycles.
     Args:
@@ -15,4 +17,5 @@ def run_clustering(cycles: list[np.array], n_clusters: int) -> KMeans:
         kmeans (KMeans):
             The fitted K-means clustering object
     """
-    return KMeans(n_clusters=n_clusters, init='random', n_init="auto", random_state=3683475120).fit(np.array(cycles))
+    return KMeans(n_clusters=n_clusters, init='random',
+                  n_init="auto" if sklearn.__version__ >= "1.2" else 10, random_state=3683475120).fit(np.array(cycles))

pypsf/hyperparameter_search.py

@@ -1,11 +1,13 @@
+from typing import Tuple
+
 import numpy as np
 from sklearn.metrics import silhouette_score, mean_absolute_error
 
 from pypsf.clustering import run_clustering
 from pypsf.predict import psf_predict
 
 
-def optimum_k(data: np.array, k_values: tuple[int]) -> int:
+def optimum_k(data: np.array, k_values: Tuple[int]) -> int:
     """
     Perform a hyperparameter search using the provided values of 'k' to
     determine the number of clusters 'k' that achieve the highest silhouette
@@ -40,7 +42,7 @@ def optimum_k(data: np.array, k_values: tuple[int]) -> int:
     return best_k
 
 
-def optimum_w(data: np.array, k: int, cycle_length: int, w_values: tuple[int]) -> int:
+def optimum_w(data: np.array, k: int, cycle_length: int, w_values: Tuple[int]) -> int:
     """
     Perform a hyperparameter search using the provided values of 'w' to
     determine the window size that results in the lowest mean absolute error.

pypsf/neighbors.py

@@ -1,9 +1,10 @@
 import re
+from typing import List
 
 import numpy as np
 
 
-def neighbor_indices(cluster_labels: np.array, w: int) -> list[int]:
+def neighbor_indices(cluster_labels: np.array, w: int) -> List[int]:
     """
     Take the last 'w' cluster labels and return all matching previous occurrences of this pattern in the rest of the
     list of cluster labels (so-called neighbors). The is done by first converting the list of ints

pypsf/psf.py

@@ -1,3 +1,5 @@
+from typing import List
+
 import numpy as np
 from numpy.typing import ArrayLike
 from sklearn.linear_model import LinearRegression
@@ -152,7 +154,7 @@ def predict(self, n_ahead: int) -> np.array:
         self.preds = self.postprocessing(preds, orig_n_ahead)
         return self.preds
 
-    def postprocessing(self, preds: list[np.array], orig_n_ahead: int) -> np.array:
+    def postprocessing(self, preds: List[np.array], orig_n_ahead: int) -> np.array:
         """
         Performs the inverse of 'preprocessing', i.e.:
             1. (Optional) Re-add a linear trend from the data if self.detrend is

pypsf/utils.py

@@ -38,6 +38,6 @@ def psf_warn(message: str):
     info = inspect.getframeinfo(caller_frame)
     orig_formatwarning = warnings.formatwarning
     warnings.formatwarning = (lambda message, category,
-                                     filename, lineno, _: f"{filename}:{lineno}:{category.__name__}:{message}\n")
+                              filename, lineno, _: f"{filename}:{lineno}:{category.__name__}:{message}\n")
     warnings.warn_explicit(message, UserWarning, info.filename, info.lineno)
-    warnings.formatwarning = orig_formatwarning
+    warnings.formatwarning = orig_formatwarning