Make tsfresh compatible with numpy 1.24

requirements.txt

@@ -1,5 +1,5 @@
 requests>=2.9.1
-numpy >=1.15.1, <1.24
+numpy >=1.15.1
 pandas>=0.25.0
 scipy>=1.2.0
 statsmodels>=0.13

tests/integrations/examples/test_robot_execution_failures.py

@@ -54,4 +54,4 @@ def test_multilabel_target_on_request(self):
  )
 
  assert len(y.unique()) > 2
- assert y.dtype == np.object
+ assert y.dtype == object

tsfresh/feature_extraction/feature_calculators.py

@@ -25,7 +25,8 @@
 
 import numpy as np
 import pandas as pd
-import stumpy
+
+# import stumpy
 from numpy.linalg import LinAlgError
 from scipy.signal import cwt, find_peaks_cwt, ricker, welch
 from scipy.stats import linregress

tsfresh/feature_selection/relevance.py

@@ -360,7 +360,7 @@ def infer_ml_task(y):
  :return: 'classification' or 'regression'
  :rtype: str
  """
- if y.dtype.kind in np.typecodes["AllInteger"] or y.dtype == np.object:
+ if y.dtype.kind in np.typecodes["AllInteger"] or y.dtype == object:
  ml_task = "classification"
  else:
  ml_task = "regression"

tsfresh/utilities/dataframe_functions.py

@@ -506,7 +506,7 @@ def roll_time_series(
 
  df = df.sort_values(column_sort)
 
- if df[column_sort].dtype != np.object:
+ if df[column_sort].dtype != object:
  # if rolling is enabled, the data should be uniformly sampled in this column
  # Build the differences between consecutive time sort values