def setUp(self):
X, y = make_blobs(n_samples=100, n_features=2, centers=2,
cluster_std=2, random_state=20)
dpp = DiProPerm(B=1000, separation_stats=['md', 't', 'auc'], clf='md')
dpp.fit(X, y)
self.dpp = dpp
self.X = X
self.y = y
def test_custom_classifier(self):
"""
User provided custom classifier.
"""
clf = get_training_fun(clf=LinearSVC(max_iter=10000),
param_grid={'C': [.0001, .001, .01, 1, 10, 100]},
metric='roc_auc', n_splits=5)
dpp = DiProPerm(B=1000, clf=clf)
dpp.fit(self.X, self.y)
self.assertTrue(hasattr(dpp, 'test_stats_'))
def test_custom_test_stats(self):
"""
User provided custom test statistics.
"""
def robust_Z(obs_stat, perm_samples):
return (obs_stat - np.median(perm_samples))/mad(perm_samples)
custom_test_stats = [('robust_Z', robust_Z)]
dpp = DiProPerm(custom_test_stats=custom_test_stats).fit(self.X, self.y)
self.assertTrue('robust_Z' in dpp.test_stats_['md'].keys())
def test_custom_sep_stats(self):
"""
User provided custom separation statistics.
"""
def median_difference(scores, y):
y = np.array(y)
classes = np.unique(y)
assert len(classes) == 2
s0 = scores[y == classes[0]]
s1 = scores[y == classes[1]]
return abs(np.median(s0) - np.median(s1))
custom_sep_stats = [('median_difference', median_difference)]
dpp = DiProPerm(custom_sep_stats=custom_sep_stats).fit(self.X, self.y)
self.assertTrue('median_difference' in dpp.test_stats_.keys())
from sklearn.datasets import make_blobs
import numpy as np
import matplotlib.pyplot as plt
from diproperm.DiProPerm import DiProPerm
X, y = make_blobs(n_samples=100,
n_features=2,
centers=2,
cluster_std=2,
random_state=20)
plt.scatter(X[:, 0], X[:, 1], c=y)
plt.savefig('data.png')
plt.close()
dpp = DiProPerm(B=1000).fit(X, y)
plt.figure(figsize=[12, 5])
# show histogram of separation statistics
plt.subplot(1, 2, 1)
dpp.plot_observed_scores()
# the observed scores
plt.subplot(1, 2, 2)
dpp.plot_perm_sep_stats(stat='md')
plt.savefig('dpp_plots.png')
plt.close()
csvReader = csv.reader(csvDataFile)
for row in csvReader:
y.append(row[1])
y = np.array(y)
X = np.array(X)
y = y[1:]
X = X[1:, 1:]
y = y.astype(int)
X = X.astype(float)
#X = pd.read_csv("full_playtime_data.csv", index_col=None, header=0, dtype = 'a')
#y = pd.read_csv("full_playtime_response.csv", index_col=None, header=0, dtype = 'a')
# DiProPerm with mean difference classifier, mean difference summary
# statistic, and 1000 permutation samples.
dpp = DiProPerm(B=1000, separation_stats=['md', 't', 'auc'], clf='md')
dpp.fit(X, y)
print(dpp.test_stats_['md'])
plt.figure(figsize=[12, 5])
# show histogram of separation statistics
plt.subplot(1, 2, 1)
dpp.plot_perm_sep_stats(stat='md')
# the observed scores
plt.subplot(1, 2, 2)
dpp.plot_observed_scores()
plt.title("all_distances")
plt.show()
def test_parallel(self):
"""
Make sure the parallel processing permutation scores runs.
"""
dpp = DiProPerm(n_jobs=-1).fit(self.X, self.y)
self.assertTrue(hasattr(dpp, 'test_stats_'))