1# Copyright 2016 The TensorFlow Authors. All Rights Reserved. 2# 3# Licensed under the Apache License, Version 2.0 (the "License"); 4# you may not use this file except in compliance with the License. 5# You may obtain a copy of the License at 6# 7# http://www.apache.org/licenses/LICENSE-2.0 8# 9# Unless required by applicable law or agreed to in writing, software 10# distributed under the License is distributed on an "AS IS" BASIS, 11# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 12# See the License for the specific language governing permissions and 13# limitations under the License. 14# ============================================================================== 15"""tf.learn IO operation tests.""" 16 17from __future__ import absolute_import 18from __future__ import division 19from __future__ import print_function 20 21import random 22 23# pylint: disable=wildcard-import 24from tensorflow.contrib.learn.python import learn 25from tensorflow.contrib.learn.python.learn import datasets 26from tensorflow.contrib.learn.python.learn.estimators._sklearn import accuracy_score 27from tensorflow.contrib.learn.python.learn.learn_io import * 28from tensorflow.python.platform import test 29 30# pylint: enable=wildcard-import 31 32 33class IOTest(test.TestCase): 34 # pylint: disable=undefined-variable 35 """tf.learn IO operation tests.""" 36 37 def test_pandas_dataframe(self): 38 if HAS_PANDAS: 39 import pandas as pd # pylint: disable=g-import-not-at-top 40 random.seed(42) 41 iris = datasets.load_iris() 42 data = pd.DataFrame(iris.data) 43 labels = pd.DataFrame(iris.target) 44 classifier = learn.LinearClassifier( 45 feature_columns=learn.infer_real_valued_columns_from_input(data), 46 n_classes=3) 47 classifier.fit(data, labels, steps=100) 48 score = accuracy_score(labels[0], list(classifier.predict_classes(data))) 49 self.assertGreater(score, 0.5, "Failed with score = {0}".format(score)) 50 else: 51 print("No pandas installed. pandas-related tests are skipped.") 52 53 def test_pandas_series(self): 54 if HAS_PANDAS: 55 import pandas as pd # pylint: disable=g-import-not-at-top 56 random.seed(42) 57 iris = datasets.load_iris() 58 data = pd.DataFrame(iris.data) 59 labels = pd.Series(iris.target) 60 classifier = learn.LinearClassifier( 61 feature_columns=learn.infer_real_valued_columns_from_input(data), 62 n_classes=3) 63 classifier.fit(data, labels, steps=100) 64 score = accuracy_score(labels, list(classifier.predict_classes(data))) 65 self.assertGreater(score, 0.5, "Failed with score = {0}".format(score)) 66 67 def test_string_data_formats(self): 68 if HAS_PANDAS: 69 import pandas as pd # pylint: disable=g-import-not-at-top 70 with self.assertRaises(ValueError): 71 learn.io.extract_pandas_data(pd.DataFrame({"Test": ["A", "B"]})) 72 with self.assertRaises(ValueError): 73 learn.io.extract_pandas_labels(pd.DataFrame({"Test": ["A", "B"]})) 74 75 def test_dask_io(self): 76 if HAS_DASK and HAS_PANDAS: 77 import pandas as pd # pylint: disable=g-import-not-at-top 78 import dask.dataframe as dd # pylint: disable=g-import-not-at-top 79 # test dask.dataframe 80 df = pd.DataFrame( 81 dict( 82 a=list("aabbcc"), b=list(range(6))), 83 index=pd.date_range( 84 start="20100101", periods=6)) 85 ddf = dd.from_pandas(df, npartitions=3) 86 extracted_ddf = extract_dask_data(ddf) 87 self.assertEqual( 88 extracted_ddf.divisions, (0, 2, 4, 6), 89 "Failed with divisions = {0}".format(extracted_ddf.divisions)) 90 self.assertEqual( 91 extracted_ddf.columns.tolist(), ["a", "b"], 92 "Failed with columns = {0}".format(extracted_ddf.columns)) 93 # test dask.series 94 labels = ddf["a"] 95 extracted_labels = extract_dask_labels(labels) 96 self.assertEqual( 97 extracted_labels.divisions, (0, 2, 4, 6), 98 "Failed with divisions = {0}".format(extracted_labels.divisions)) 99 # labels should only have one column 100 with self.assertRaises(ValueError): 101 extract_dask_labels(ddf) 102 else: 103 print("No dask installed. dask-related tests are skipped.") 104 105 def test_dask_iris_classification(self): 106 if HAS_DASK and HAS_PANDAS: 107 import pandas as pd # pylint: disable=g-import-not-at-top 108 import dask.dataframe as dd # pylint: disable=g-import-not-at-top 109 random.seed(42) 110 iris = datasets.load_iris() 111 data = pd.DataFrame(iris.data) 112 data = dd.from_pandas(data, npartitions=2) 113 labels = pd.DataFrame(iris.target) 114 labels = dd.from_pandas(labels, npartitions=2) 115 classifier = learn.LinearClassifier( 116 feature_columns=learn.infer_real_valued_columns_from_input(data), 117 n_classes=3) 118 classifier.fit(data, labels, steps=100) 119 predictions = data.map_partitions(classifier.predict).compute() 120 score = accuracy_score(labels.compute(), predictions) 121 self.assertGreater(score, 0.5, "Failed with score = {0}".format(score)) 122 123 124if __name__ == "__main__": 125 test.main() 126