Updated libraries to newest versions.

Stability fixes.

EvoPreprocess/__init__.py

@@ -18,5 +18,5 @@
 
 name = 'EvoPreprocess'
 __project__ = 'EvoPreprocess'
-__version__ = '0.3.1'
+__version__ = '0.3.2'
 __all__ = ['feature_selection', 'data_sampling', 'data_weighting']

EvoPreprocess/data_sampling/SamplingBenchmark.py

@@ -10,12 +10,11 @@
 import random
 
 import numpy as np
-import pandas as pd
 from NiaPy.benchmarks import Benchmark
 from sklearn.base import ClassifierMixin
 from sklearn.metrics import mean_squared_error, f1_score
 from sklearn.naive_bayes import GaussianNB
-from sklearn.utils import safe_indexing
+from sklearn.utils import safe_indexing, check_X_y
 
 
 class SamplingBenchmark(Benchmark):
@@ -56,10 +55,7 @@ def __init__(self,
         self.Upper = 1
         super().__init__(self.Lower, self.Upper)
 
-        if isinstance(X, pd.DataFrame):
-            X = X.values
-        if isinstance(y, pd.Series):
-            y = y.values
+        X, y = check_X_y(X, y, force_all_finite=False)
 
         self.X_train, self.X_valid = X[train_indices, :], X[valid_indices, :]
         self.y_train, self.y_valid = y[train_indices], y[valid_indices]
@@ -97,7 +93,10 @@ def to_phenotype(genotype):
         setting = np.cumsum(genotype[-5:])
         appearances = genotype[:-5]
 
-        return np.digitize(appearances, setting / setting[-1])
+        if setting[-1] == 0:
+            return np.digitize(appearances, setting)
+        else:
+            return np.digitize(appearances, setting / setting[-1])
 
     @staticmethod
     def genotype_to_map(genotype):

EvoPreprocess/data_weighting/WeightingBenchmark.py

@@ -7,11 +7,11 @@
 # License: GNU General Public License v3.0
 import random
 
-import pandas as pd
 from NiaPy.benchmarks import Benchmark
 from sklearn.base import ClassifierMixin
 from sklearn.metrics import accuracy_score, mean_squared_error
 from sklearn.naive_bayes import GaussianNB
+from sklearn.utils import check_X_y
 
 
 class WeightingBenchmark(Benchmark):
@@ -48,10 +48,7 @@ def __init__(self,
         self.Upper = 2  # Max weight of the instance, should be tested
         super().__init__(self.Lower, self.Upper)
 
-        if isinstance(X, pd.DataFrame):
-            X = X.values
-        if isinstance(y, pd.Series):
-            y = y.values
+        X, y = check_X_y(X, y, force_all_finite=False)
 
         self.X_train, self.X_valid = X[train_indices], X[valid_indices]
         self.y_train, self.y_valid = y[train_indices], y[valid_indices]
@@ -65,10 +62,17 @@ def __init__(self,
 
     def function(self):
         def evaluate(D, sol):
-            cls = self.evaluator.fit(self.X_train, self.y_train, sample_weight=sol)
+            weights = WeightingBenchmark.to_phenotype(sol, self.Upper)
+            cls = self.evaluator.fit(self.X_train, self.y_train, sample_weight=weights)
             y_predicted = cls.predict(self.X_valid)
             acc = self.metric(self.y_valid, y_predicted)
             acc = (1 - acc) if self.evaluator is ClassifierMixin else acc
             return acc
 
         return evaluate
+
+    @staticmethod
+    def to_phenotype(sol, Upper):
+        max_value = sol[-1] / Upper
+        weights = sol[:-1]
+        return weights * max_value

EvoPreprocess/feature_selection/FeatureSelectionBenchmark.py

@@ -10,12 +10,12 @@
 import random
 
 import numpy as np
-import pandas as pd
 from NiaPy.benchmarks import Benchmark
 from sklearn.base import ClassifierMixin
 from sklearn.metrics import f1_score
 from sklearn.metrics import mean_squared_error
 from sklearn.naive_bayes import GaussianNB
+from sklearn.utils import check_X_y
 
 
 class FeatureSelectionBenchmark(Benchmark):
@@ -58,10 +58,7 @@ def __init__(self,
         super().__init__(self.Lower, self.Upper)
 
         self.split = split
-        if isinstance(X, pd.DataFrame):
-            X = X.values
-        if isinstance(y, pd.Series):
-            y = y.values
+        X, y = check_X_y(X, y, force_all_finite=False)
 
         self.X_train, self.X_valid = X[train_indices, :], X[valid_indices, :]
         self.y_train, self.y_valid = y[train_indices], y[valid_indices]

Examples/ExampleSamplingCustomOptimizer.py

@@ -13,9 +13,14 @@ class RandomSearch(Algorithm):
     Name = ['RandomSearch', 'RS']
 
     def runIteration(self, task, pop, fpop, xb, fxb, **dparams):
-        pop = task.Lower + self.Rand.rand(self.NP, task.D) * task.bRange
-        fpop = apply_along_axis(task.eval, 1, pop)
-        return pop, fpop, {}
+        try:
+            pop = task.Lower + self.Rand.rand(self.NP, task.D) * task.bRange
+            fpop = apply_along_axis(task.eval, 1, pop)
+            xb, fxb = self.getBest(pop, fpop, xb, fxb)
+            return pop, fpop, xb, fxb, {}
+        except Exception as x:
+            print(x)
+            return None, None, None
 
 
 class CustomSamplingBenchmark(SamplingBenchmark):
@@ -24,8 +29,7 @@ class CustomSamplingBenchmark(SamplingBenchmark):
 
     def function(self):
         def evaluate(D, sol):
-            phenotype = SamplingBenchmark.map_to_phenotype(
-                CustomSamplingBenchmark.to_phenotype(sol))
+            phenotype = SamplingBenchmark.map_to_phenotype(CustomSamplingBenchmark.to_phenotype(sol))
             X_sampled = safe_indexing(self.X_train, phenotype)
             y_sampled = safe_indexing(self.y_train, phenotype)
 

README.md

@@ -13,12 +13,11 @@ These instructions will get you a copy of the project up and running on your loc
 ### Dependencies
 EvoSampling requires:
 
-- numpy(>=1.8.2)
-- scipy(>=0.17)
-- pandas
-- scikit-learn(>=0.19.0)
-- imbalanced-learn(>=0.3.1)
-- NiaPy(>=2.0.0rc5)
+- numpy
+- scipy
+- scikit-learn
+- imbalanced-learn
+- NiaPy(>=2.0.0rc11)
 
 ### Installation
 Install EvoPreprocess with pip:

requirements.txt

@@ -1,6 +1,5 @@
-numpy>=1.8.2
-pandas
-scipy>=0.17
-scikit-learn>=0.19.0
-imbalanced-learn>=0.3.1
-NiaPy>=2.0.0rc5
+numpy
+scipy
+scikit-learn
+imbalanced-learn
+NiaPy>=2.0.0rc11

setup.py

@@ -49,12 +49,11 @@ def read_package_variable(key, filename='__init__.py'):
         'Topic :: Scientific/Engineering :: Information Analysis'
     ],
     install_requires=[
-        'numpy>=1.8.2',
-        'pandas',
-        'scipy>=0.17',
-        'scikit-learn>=0.19.0'
-        'imbalanced-learn>=0.3.1'
-        'NiaPy>=2.0.0rc5'
+        'numpy',
+        'scipy',
+        'scikit-learn'
+        'imbalanced-learn'
+        'NiaPy>=2.0.0rc11'
     ],
     keywords=[
         'Evolutionary Algorithms',