Major refactoring

test/Visualization.py

@@ -0,0 +1,141 @@
+from __future__ import division
+from __future__ import print_function
+
+__author__ = 'fnogueira'
+
+s = {
+ "lines.linewidth": 2.0,
+ "patch.linewidth": 0.5,
+ "legend.fancybox": True,
+ "axes.color_cycle": [
+ "#30a2da",
+ "#fc4f30",
+ "#e5ae38",
+ "#6d904f",
+ "#8b8b8b"
+ ],
+ "axes.facecolor": "#f0f0f0",
+ "axes.labelsize": "large",
+ "axes.axisbelow": True,
+ "axes.grid": True,
+ "patch.edgecolor": "#f0f0f0",
+ "axes.titlesize": "x-large",
+ "svg.embed_char_paths": "path",
+ "examples.directory": "",
+ "figure.facecolor": "#f0f0f0",
+ "grid.linestyle": "-",
+ "grid.linewidth": 1.0,
+ "grid.color": "#cbcbcb",
+ "axes.edgecolor":"#f0f0f0",
+ "xtick.major.size": 0,
+ "xtick.minor.size": 0,
+ "ytick.major.size": 0,
+ "ytick.minor.size": 0,
+ "axes.linewidth": 3.0,
+ "font.size": 14.0,
+ "lines.solid_capstyle": "butt",
+ "savefig.edgecolor": "#f0f0f0",
+ "savefig.facecolor": "#f0f0f0",
+ "figure.subplot.left" : 0.08,
+ "figure.subplot.right" : 0.95,
+ "figure.subplot.bottom" : 0.07
+}
+
+def vizualization():
+
+ from sklearn.datasets import make_classification
+ x, y = make_classification(n_classes=2, class_sep=2, weights=[0.1, 0.9],\
+ n_informative=3, n_redundant=1, flip_y=0,\
+ n_features=20, n_clusters_per_class=1,\
+ n_samples=5000, random_state=10)
+
+
+
+ from UnbalancedDataset import OverSampler, SMOTE, bSMOTE1, bSMOTE2, SVM_SMOTE
+
+ ratio = 1
+ new = ratio * 500
+
+ # -------------------------------- // -------------------------------- #
+ # Datasets
+ OS = OverSampler(random_state=1)
+ ox, oy = OS.fit_transform(x, y)
+
+ smote = SMOTE(random_state=1)
+ sx, sy = smote.fit_transform(x, y)
+
+ bsmote1 = bSMOTE1(random_state=1)
+ bsx1, bsy1 = bsmote1.fit_transform(x, y)
+
+ bsmote2 = bSMOTE2(random_state=1)
+ bsx2, bsy2 = bsmote2.fit_transform(x, y)
+
+ svm_args={'class_weight' : 'auto'}
+ svmsmote = SVM_SMOTE(random_state=1, **svm_args)
+ svmx, svmy = svmsmote.fit_transform(x, y)
+
+
+ # -------------------------------- // -------------------------------- #
+ # PCA
+ from sklearn.decomposition import PCA
+ pca = PCA(n_components = 2)
+
+ x = pca.fit_transform(x)
+
+ ox = pca.transform(ox)
+ sx = pca.transform(sx)
+ bsx1 = pca.transform(bsx1)
+ bsx2 = pca.transform(bsx2)
+ svmx = pca.transform(svmx)
+
+ # -------------------------------- // -------------------------------- #
+ # Visualization
+ import matplotlib
+ matplotlib.rcParams.update(s)
+ import matplotlib.pyplot as plt
+
+ f, ax = plt.subplots(2, 3, figsize=(16, 9))
+ f.suptitle("Over-sampling with SMOTE: comparison", fontsize=16)
+
+ for e, c in zip(set(y), ['purple', 'g']):
+ ax[0, 0].scatter(x[y==e, 0], x[y==e, 1], color = c, alpha = 0.5)
+ ax[0, 0].set_title('Original', fontsize=12)
+
+ for e, c in zip(set(y), ['purple', 'g']):
+ ax[0, 1].scatter(x[y==e, 0], x[y==e, 1], color = c, alpha = 0.5)
+ ax[0, 1].scatter(ox[-new:, 0], ox[-new:, 1], color = 'y', alpha = 0.3)
+ ax[0, 1].set_title('Random Over-sampling', fontsize=12)
+
+
+ for e, c in zip(set(y), ['purple', 'g']):
+ ax[0, 2].scatter(x[y==e, 0], x[y==e, 1], color = c, alpha = 0.5)
+ ax[0, 2].scatter(sx[-new:, 0], sx[-new:, 1], color = 'y', alpha = 0.3)
+ ax[0, 2].set_title('SMOTE', fontsize=12)
+
+
+ for e, c in zip(set(y), ['purple', 'g']):
+ ax[1, 0].scatter(x[y==e, 0], x[y==e, 1], color = c, alpha = 0.5)
+ ax[1, 0].scatter(bsx1[-new:, 0], bsx1[-new:, 1], color = 'y', alpha = 0.3)
+ ax[1, 0].set_title('Borderline-SMOTE type 1', fontsize=12)
+
+ for e, c in zip(set(y), ['purple', 'g']):
+ ax[1, 1].scatter(x[y==e, 0], x[y==e, 1], color = c, alpha = 0.5)
+ ax[1, 1].scatter(bsx2[-new:, 0], bsx2[-new:, 1], color = 'y', alpha = 0.3)
+ ax[1, 1].set_title('Borderline-SMOTE type 2', fontsize=12)
+
+ for e, c in zip(set(y), ['purple', 'g']):
+ ax[1, 2].scatter(x[y==e, 0], x[y==e, 1], color = c, alpha = 0.5)
+ ax[1, 2].scatter(svmx[-new:, 0], svmx[-new:, 1], color = 'y', alpha = 0.3)
+ ax[1, 2].set_title('SVM-SMOTE', fontsize=12)
+
+ # Hide ticks
+ plt.setp([a.get_xticklabels() for a in ax[0, :]], visible=False)
+ plt.setp([a.get_yticklabels() for a in ax[:, 1]], visible=False)
+ plt.setp([a.get_yticklabels() for a in ax[:, 2]], visible=False)
+
+ plt.show()
+
+if __name__ == '__main__':
+
+ vizualization()
+

unbalanced_dataset/__init__.py

@@ -0,0 +1,31 @@
+"""
+The :mod:`UnbalancedDataset` module includes methods for
+tackling the problem of inbalanced datasets.
+"""
+
+from .unbalanced_dataset import UnbalancedDataset
+
+from .over_sampling import OverSampler
+from .over_sampling import SMOTE
+
+from .under_sampling import UnderSampler
+from .under_sampling import TomekLinks
+from .under_sampling import ClusterCentroids
+from .under_sampling import NearMiss
+from .under_sampling import CondensedNearestNeighbour
+from .under_sampling import OneSidedSelection
+from .under_sampling import NeighbourhoodCleaningRule
+
+from .ensemble_sampling import EasyEnsemble
+from .ensemble_sampling import BalanceCascade
+
+from pipeline import SMOTEENN
+from pipeline import SMOTETomek
+from pipeline import Pipeline
+
+__all__ = ["UnderSampler", "NearMiss",
+ "CondensedNearestNeighbour", "OneSidedSelection",
+ "NeighbourhoodCleaningRule", "TomekLinks",
+ "ClusterCentroids", "OverSampler", "SMOTE", "SMOTETomek",
+ "SMOTEENN", "EasyEnsemble", "BalanceCascade"]
+