upgrade libs versions

Author: DanilZherebtsov

README.rst

@@ -1,7 +1,7 @@
 .. image:: https://img.shields.io/pepy/dt/verstack
    :target: https://pypi.org/project/verstack/
 
-.. image:: https://img.shields.io/badge/version-4.3.3-success.svg?color=blue
+.. image:: https://img.shields.io/badge/version-4.3.4-success.svg?color=blue
    :target: https://pypi.org/project/verstack/
 
 .. image:: logo.png

docs/source/index.rst

@@ -1,5 +1,5 @@
 ############################
-verstack 4.3.3 Documentation
+verstack 4.3.4 Documentation
 ############################
 Machine learning tools to make a Data Scientist's work efficient
 
@@ -1115,7 +1115,7 @@ Impute all missing values in a pandas dataframe by xgboost models in multiproces
 
 .. note:: 
   This is the second major version of `NaNImputer`. The original class (last version 1.4.0) 
-  had been very popular. The legacy version is `NaNImputerLegacy` has been depresciated after verstack version 4.3.3
+  had been very popular. The legacy version is `NaNImputerLegacy` has been depresciated after verstack version 4.3.4
   Differences between the NaNImputer and NaNImputerLegacy:
       - The new NaNImputer is based on LightGBM instead of XGBoost in the legacy version
       - The new NaNImputer is using a single core multithreading instead of multicore legacy version, it is significantly faster

requirements.txt

@@ -1,17 +1,17 @@
-numpy>=1.26.4,<=2.1.1
-pandas==2.2.2
-scikit-learn>=1.3.2,<=1.5.1
-lightgbm>=4.4.0,<=4.5.0
-optuna>=3.5.0,<=4.0.0
+numpy~=2.3.0
+pandas~=2.3.0
+scikit-learn~=1.7.1
+lightgbm~=4.5.0
+optuna~=4.0.0
 optuna-integration>=3.2.0,<=4.0.0
-plotly>=5.11.0,<=5.24.0
-matplotlib==3.9.2
-seaborn==0.13.2
-python-dateutil==2.9.0
-holidays==0.56
-mlxtend==0.23.1
-category_encoders>=2.5.1,<=2.6.3
-tabulate==0.9.0
+plotly~=5.24.0
+matplotlib~=3.9.2
+seaborn~=0.13.2
+python-dateutil~=2.9.0
+holidays~=0.56
+mlxtend~=0.23.1
+category_encoders~=2.6.3
+tabulate~=0.9.0
 # following packages for testing purposes only
-# fastparquet==2024.5.0
-# pytest==8.3.2
+# fastparquet~=2024.5.0
+# pytest~=8.3.2

verstack/categoric_encoders/WeightOfEvidenceEncoder.py

@@ -1,5 +1,6 @@
 import pandas as pd
 import numpy as np
+from category_encoders import WOEEncoder
 from verstack.categoric_encoders.args_validators import is_bool_na_sentinel, assert_fit_transform_args, assert_transform_args, assert_binary_target
 from verstack.tools import Printer
 
@@ -81,7 +82,6 @@ def fit_transform(self, df, colname, targetname):
         assert_binary_target(df, targetname)
         encoded_df = df.copy()
         self._colname = colname
-        from category_encoders import WOEEncoder
         generic_encoder = WOEEncoder(**self._params)
         encoded_column = generic_encoder.fit_transform(df[colname], df[targetname])
         self.__generic_encoder = generic_encoder

verstack/lgbm_optuna_tuning/lgb_metrics.py

@@ -8,7 +8,9 @@
 
 import numpy as np
 from sklearn.metrics import mean_absolute_error as mae
+from sklearn.metrics import root_mean_squared_error as rmse
 from sklearn.metrics import mean_squared_error as mse
+from sklearn.metrics import root_mean_squared_log_error as rmsle
 from sklearn.metrics import mean_squared_log_error as msle
 from sklearn.metrics import r2_score as r2
 from sklearn.metrics import roc_auc_score as auc
@@ -86,17 +88,24 @@ def _percentage_error(real, pred):
 # mae imported from sklearn
 # ------------------------------------------------------------------------------
 
-def rmse(real, pred):
-    return mse(real, pred, squared = False)
+# rmse imported from sklearn
 # ------------------------------------------------------------------------------
 
-def rmsle(real, pred):
-    '''Changes negative predictions to 0 for correct calculation'''
-    try:
-        return msle(real, pred, squared = False)
-    except ValueError:
-        pred_non_negative, real_non_negative = remove_negatives(pred, real)
-        return msle(real_non_negative, pred_non_negative, squared = False)
+# rmsle imported from sklearn
+# ------------------------------------------------------------------------------
+
+# def rmse(real, pred):
+#     return mse(real, pred, squared = False)
+# ------------------------------------------------------------------------------
+
+# def rmsle(real, pred):
+#     '''Changes negative predictions to 0 for correct calculation'''
+#     try:
+#         return msle(real, pred, squared = False)
+#     except ValueError:
+#         pred_non_negative, real_non_negative = remove_negatives(pred, real)
+#         return msle(real_non_negative, pred_non_negative, squared = False)
+
 
 # ------------------------------------------------------------------------------
 

verstack/tests/conftest.py

@@ -0,0 +1,26 @@
+import sys
+import os
+
+# Get the verstack directory (parent of tests)
+verstack_dir = os.path.dirname(os.path.dirname(__file__))
+tests_dir = os.path.dirname(__file__)
+
+# Remove any existing verstack-related paths from sys.path
+sys.path = [p for p in sys.path if not (
+    'verstack' in p and ('site-packages' in p or 'dist-packages' in p)
+)]
+
+# Insert local paths at the very beginning
+sys.path.insert(0, verstack_dir)
+sys.path.insert(0, tests_dir)
+
+print(f"Verstack dir: {verstack_dir}")
+print(f"Tests dir: {tests_dir}")
+print(f"Cleaned sys.path first 5: {sys.path[:5]}")
+
+# Debug: Check what's in the directories
+print(f"Contents of verstack_dir: {os.listdir(verstack_dir)}")
+if os.path.exists(os.path.join(verstack_dir, 'lgbm_optuna_tuning')):
+    print(f"lgbm_optuna_tuning exists: {os.path.exists(os.path.join(verstack_dir, 'lgbm_optuna_tuning', 'LGBMTuner.py'))}")
+if os.path.exists(os.path.join(verstack_dir, 'categoric_encoders')):
+    print(f"categoric_encoders exists: {os.path.exists(os.path.join(verstack_dir, 'categoric_encoders', 'WeightOfEvidenceEncoder.py'))}")

verstack/tests/test_DateParser.py

@@ -1,30 +1,31 @@
-import sys
 import numpy as np
 import pandas as pd
+import os
 
-sys.path.append("../")
 from DateParser import DateParser
 
+# Get the directory where this test file is located
+test_dir = os.path.dirname(__file__)
+
 datasets = {
     1: {
-        "train": "dateparser_train_1.parquet",
-        "test": "dateparser_test_1.parquet",
+        "train": os.path.join(test_dir, "dateparser_train_1.parquet"),
+        "test": os.path.join(test_dir, "dateparser_test_1.parquet"),
     },
     2: {
-        "train": "dateparser_train_2.parquet",
-        "test": "dateparser_test_2.parquet",
+        "train": os.path.join(test_dir, "dateparser_train_2.parquet"),
+        "test": os.path.join(test_dir, "dateparser_test_2.parquet"),
     },
     3: {
-        "train": "dateparser_train_3.parquet",
-        "test": "dateparser_test_3.parquet",
+        "train": os.path.join(test_dir, "dateparser_train_3.parquet"),
+        "test": os.path.join(test_dir, "dateparser_test_3.parquet"),
     },
     4: {
-        "train": "dateparser_train_4.parquet",
-        "test": "dateparser_test_4.parquet",
+        "train": os.path.join(test_dir, "dateparser_train_4.parquet"),
+        "test": os.path.join(test_dir, "dateparser_test_4.parquet"),
     },
 }
 
-
 # test overall DateParser not being broken
 def test_DateParser():
     result = []
@@ -37,4 +38,4 @@ def test_DateParser():
         result.append(
             np.all(transformed_train.columns == transformed_test.columns)
         )
-    assert result
+    assert result

verstack/tests/test_DateParser_new.py

@@ -1,8 +1,4 @@
 import pandas as pd
-import numpy as np
-import sys
-
-sys.path.append("../")
 from DateParser import DateParser
 
 def test_all_date_formats_and_transform():

verstack/tests/test_Factorizer.py

@@ -1,5 +1,3 @@
-import sys
-sys.path.append("../")
 import numpy as np
 from common import generate_data
 from categoric_encoders.Factorizer import Factorizer

verstack/tests/test_FrequencyEncoder.py

@@ -1,5 +1,3 @@
-import sys
-sys.path.append("../")
 import numpy as np
 from common import generate_data
 from categoric_encoders.FrequencyEncoder import FrequencyEncoder