def test_filelike(self):
"""Test reading from file-like object (StringIO)"""
f1 = open(test1)
f2 = StringIO(open(test1).read())
data1, meta1 = loadarff(f1)
data2, meta2 = loadarff(f2)
assert_(data1 == data2)
assert_(repr(meta1) == repr(meta2))
def test_filelike(self):
"""Test reading from file-like object (StringIO)"""
f1 = open(test1)
f2 = StringIO(open(test1).read())
data1, meta1 = loadarff(f1)
data2, meta2 = loadarff(f2)
assert_(data1 == data2)
assert_(repr(meta1) == repr(meta2))
def test_filelike(self):
# Test reading from file-like object (StringIO)
with open(test1) as f1:
data1, meta1 = loadarff(f1)
with open(test1) as f2:
data2, meta2 = loadarff(StringIO(f2.read()))
assert_(data1 == data2)
assert_(repr(meta1) == repr(meta2))
def test_filelike(self):
# Test reading from file-like object (StringIO)
f1 = open(test1)
data1, meta1 = loadarff(f1)
f1.close()
f2 = open(test1)
data2, meta2 = loadarff(StringIO(f2.read()))
f2.close()
assert_(data1 == data2)
assert_(repr(meta1) == repr(meta2))
def test_filelike(self):
# Test reading from file-like object (StringIO)
f1 = open(test1)
data1, meta1 = loadarff(f1)
f1.close()
f2 = open(test1)
data2, meta2 = loadarff(StringIO(f2.read()))
f2.close()
assert_(data1 == data2)
assert_(repr(meta1) == repr(meta2))
def test_path(self):
# Test reading from `pathlib.Path` object
from pathlib import Path
with open(test1) as f1:
data1, meta1 = loadarff(f1)
data2, meta2 = loadarff(Path(test1))
assert_(data1 == data2)
assert_(repr(meta1) == repr(meta2))
def test_path(self):
# Test reading from `pathlib.Path` object
from pathlib import Path
with open(test1) as f1:
data1, meta1 = loadarff(f1)
data2, meta2 = loadarff(Path(test1))
assert_(data1 == data2)
assert_(repr(meta1) == repr(meta2))
def test_nodata(self):
# The file nodata.arff has no data in the @DATA section.
# Reading it should result in an array with length 0.
nodata_filename = os.path.join(data_path, 'nodata.arff')
data, meta = loadarff(nodata_filename)
expected_dtype = np.dtype([
('sepallength', '<f8'), ('sepalwidth', '<f8'),
('petallength', '<f8'), ('petalwidth', '<f8'), ('class', 'S15')
])
assert_equal(data.dtype, expected_dtype)
assert_equal(data.size, 0)
def test_nodata(self):
# The file nodata.arff has no data in the @DATA section.
# Reading it should result in an array with length 0.
nodata_filename = os.path.join(data_path, 'nodata.arff')
data, meta = loadarff(nodata_filename)
expected_dtype = np.dtype([('sepallength', '<f8'),
('sepalwidth', '<f8'),
('petallength', '<f8'),
('petalwidth', '<f8'),
('class', 'S15')])
assert_equal(data.dtype, expected_dtype)
assert_equal(data.size, 0)
def characterize(file_name):
a = arffread.loadarff(file_name)
df = pd.DataFrame(a[0])
df = numerify_columns(df)
X, y = df.loc[:, df.columns != 'LEAVE'], df['LEAVE']
X_train, X_test, y_train, y_test = train_test_split(X,
y,
test_size=0.3,
train_size=0.5)
decision_tree(X_train, y_train)
min_cols, max_cols = 3, len(X_train.columns)
for evaluator in [evaluate_by_rank, evaluate_by_pca]:
best_evaluated = best_ranker_accuracy(X_train, y_train, evaluator,
min_cols, max_cols)
log.info(f"Best {evaluator}: {best_evaluated}")
def test_missing(self):
data, meta = loadarff(missing)
for i in ['yop', 'yap']:
assert_array_almost_equal(data[i], expect_missing[i])
def _test(self, test_file):
data, meta = loadarff(test_file)
for i in range(len(data)):
for j in range(4):
assert_array_almost_equal(expect4_data[i][j], data[i][j])
assert_equal(meta.types(), expected_types)
def setup_method(self):
self.data, self.meta = loadarff(test_quoted_nominal_spaces)
def setup_method(self):
self.data, self.meta = loadarff(test10)
def setup_method(self):
self.data, self.meta = loadarff(test10)
def setUp(self):
self.data, self.meta = loadarff(test7)
def setUp(self):
self.data, self.meta = loadarff(test7)
def _test(self, test_file):
data, meta = loadarff(test_file)
for i in range(len(data)):
for j in range(4):
assert_array_almost_equal(expect4_data[i][j], data[i][j])
assert_equal(meta.types(), expected_types)
def test_missing(self):
data, meta = loadarff(missing)
for i in ['yop', 'yap']:
assert_array_almost_equal(data[i], expect_missing[i])
import pandas as pd
import numpy as np
from scipy.io.arff import arffread
from sklearn.tree import DecisionTreeClassifier
from sklearn.model_selection import train_test_split, RepeatedStratifiedKFold, cross_val_score
from sklearn.feature_selection import RFE
from sklearn.pipeline import Pipeline
if __name__ == '__main__':
a = arffread.loadarff('../data/churn.arff')
churn = pd.DataFrame(a[0])
type_map = {}
for c in churn.columns:
if churn[c].dtype.name == 'object':
churn[c] = churn[c].apply(lambda x: x.decode('utf8'))
type_map[c] = ['empty'] + list(churn[c].unique())
churn.loc[churn[c].isna(), c] = type_map[c][0]
churn[c] = churn[c].apply(lambda l: type_map[c].index(l))
churn[c].astype(int)
X, y = churn.loc[:, churn.columns != 'LEAVE'], churn['LEAVE']
X_train, X_test, y_train, y_test = train_test_split(X,
y,
test_size=0.3,
train_size=0.5)
classer = DecisionTreeClassifier()
print(f"churn columns: {churn.columns}")
result = classer.fit(X_train, y_train)
print(f"CLASSER RESULT: {result}")
important_cols = [
c for i, c in enumerate(X.columns)
if result.feature_importances_[i] > 0.07
