1.0.1 release: python 3.7 support

oats/generator/_base.py

@@ -1,8 +1,8 @@
-from typing import Protocol, Tuple
+from typing import Tuple
 import numpy.typing as npt
 
 
-class Generator(Protocol):
+class Generator():
     """Base class for Generators
     Because of complex behavior of generators. Method signatures are not strictly enforced.
     But must implement `get_dataset()` method that returns `(train, test, label)`

oats/models/_base.py

@@ -1,9 +1,8 @@
-from typing import Protocol
 import numpy.typing as npt
 import numpy as np
 
 
-class Model(Protocol):
+class Model():
     """Base class for Models
 
     Preprocessors are any function that transforms timeseries T to T'.

oats/models/predictive/rnn.py

@@ -2,7 +2,7 @@
 Recurrent Neural Networks (RNN)
 -----------------
 """
-from typing import Any, Literal
+from typing import Any
 from functools import partial
 
 from darts import models

oats/models/rule_based/quantile.py

@@ -40,7 +40,6 @@ def get_scores(self, data):
                 window_threshold = np.percentile(
                     data[i - self.window : i],
                     self.quantile * 100,
-                    method="closest_observation",
                 )
                 E[i] = 1 if data[i] > window_threshold else 0
 

oats/preprocessor/_base.py

@@ -1,9 +1,7 @@
-from typing import Protocol
-
 import numpy as np
 
 
-class Preprocessor(Protocol):
+class Preprocessor():
     """Base class for Preprocessors.
 
     Preprocessors are any function that transforms timeseries T to T'.

oats/scorer/_base.py

@@ -1,7 +1,4 @@
-from typing import Protocol
-
-
-class Scorer(Protocol):
+class Scorer():
     """Base class for Scorers.
     Scorer computes the result of a predictor; can be either unsupervised or supervised.
 

oats/threshold/_base.py

@@ -1,9 +1,9 @@
-from typing import Protocol, Any
+from typing import Any
 
 import numpy as np
 
 
-class Threshold(Protocol):
+class Threshold():
     """Base class for thresholders
 
     Preprocessor object must take a `fit()` method as well as `get_threshold()`.