Python* API Reference for Intel® Data Analytics Acceleration Library 2018 Update 1

dt_cls_dense_batch.py

1 #===============================================================================
2 # Copyright 2014-2017 Intel Corporation
3 # All Rights Reserved.
4 #
5 # If this software was obtained under the Intel Simplified Software License,
6 # the following terms apply:
7 #
8 # The source code, information and material ("Material") contained herein is
9 # owned by Intel Corporation or its suppliers or licensors, and title to such
10 # Material remains with Intel Corporation or its suppliers or licensors. The
11 # Material contains proprietary information of Intel or its suppliers and
12 # licensors. The Material is protected by worldwide copyright laws and treaty
13 # provisions. No part of the Material may be used, copied, reproduced,
14 # modified, published, uploaded, posted, transmitted, distributed or disclosed
15 # in any way without Intel's prior express written permission. No license under
16 # any patent, copyright or other intellectual property rights in the Material
17 # is granted to or conferred upon you, either expressly, by implication,
18 # inducement, estoppel or otherwise. Any license under such intellectual
19 # property rights must be express and approved by Intel in writing.
20 #
21 # Unless otherwise agreed by Intel in writing, you may not remove or alter this
22 # notice or any other notice embedded in Materials by Intel or Intel's
23 # suppliers or licensors in any way.
24 #
25 #
26 # If this software was obtained under the Apache License, Version 2.0 (the
27 # "License"), the following terms apply:
28 #
29 # You may not use this file except in compliance with the License. You may
30 # obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0
31 #
32 #
33 # Unless required by applicable law or agreed to in writing, software
34 # distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
35 # WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
36 #
37 # See the License for the specific language governing permissions and
38 # limitations under the License.
39 #===============================================================================
40 
41 
42 
43 
44 import os
45 import sys
46 
47 from daal.algorithms.decision_tree.classification import prediction, training
48 from daal.algorithms import classifier
49 from daal.data_management import (
50  FileDataSource, DataSourceIface, NumericTableIface, HomogenNumericTable, MergedNumericTable
51 )
52 utils_folder = os.path.realpath(os.path.abspath(os.path.dirname(os.path.dirname(__file__))))
53 if utils_folder not in sys.path:
54  sys.path.insert(0, utils_folder)
55 from utils import printNumericTables
56 
57 DAAL_PREFIX = os.path.join('..', 'data')
58 
59 # Input data set parameters
60 trainDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'decision_tree_train.csv')
61 pruneDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'decision_tree_prune.csv')
62 testDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'decision_tree_test.csv')
63 
64 nFeatures = 5
65 nClasses = 5
66 
67 # Model object for the decision tree classification algorithm
68 model = None
69 predictionResult = None
70 testGroundTruth = None
71 
72 
73 def trainModel():
74  global model
75 
76  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
77  trainDataSource = FileDataSource(
78  trainDatasetFileName,
79  DataSourceIface.notAllocateNumericTable,
80  DataSourceIface.doDictionaryFromContext
81  )
82 
83  # Create Numeric Tables for training data and labels
84  trainData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
85  trainGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
86  mergedData = MergedNumericTable(trainData, trainGroundTruth)
87 
88  # Retrieve the data from the input file
89  trainDataSource.loadDataBlock(mergedData)
90 
91  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
92  pruneDataSource = FileDataSource(
93  pruneDatasetFileName,
94  DataSourceIface.notAllocateNumericTable,
95  DataSourceIface.doDictionaryFromContext
96  )
97 
98  # Create Numeric Tables for pruning data and labels
99  pruneData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
100  pruneGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
101  pruneMergedData = MergedNumericTable(pruneData, pruneGroundTruth)
102 
103  # Retrieve the data from the input file
104  pruneDataSource.loadDataBlock(pruneMergedData)
105 
106  # Create an algorithm object to train the decision tree classification model
107  algorithm = training.Batch(nClasses)
108 
109  # Pass the training data set and dependent values to the algorithm
110  algorithm.input.set(classifier.training.data, trainData)
111  algorithm.input.set(classifier.training.labels, trainGroundTruth)
112  algorithm.input.setTable(training.dataForPruning, pruneData)
113  algorithm.input.setTable(training.labelsForPruning, pruneGroundTruth)
114 
115  # Train the decision tree classification model and retrieve the results of the training algorithm
116  trainingResult = algorithm.compute()
117  model = trainingResult.get(classifier.training.model)
118 
119 def testModel():
120  global testGroundTruth, predictionResult
121 
122  # Initialize FileDataSource<CSVFeatureManager> to retrieve the test data from a .csv file
123  testDataSource = FileDataSource(
124  testDatasetFileName,
125  DataSourceIface.notAllocateNumericTable,
126  DataSourceIface.doDictionaryFromContext
127  )
128 
129  # Create Numeric Tables for testing data and labels
130  testData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
131  testGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
132  mergedData = MergedNumericTable(testData, testGroundTruth)
133 
134  # Retrieve the data from input file
135  testDataSource.loadDataBlock(mergedData)
136 
137  # Create algorithm objects for decision tree classification prediction with the default method
138  algorithm = prediction.Batch()
139 
140  # Pass the testing data set and trained model to the algorithm
141  print("Number of columns: {}".format(testData.getNumberOfColumns()))
142  algorithm.input.setTable(classifier.prediction.data, testData)
143  algorithm.input.setModel(classifier.prediction.model, model)
144 
145  # Compute prediction results and retrieve algorithm results
146  # (Result class from classifier.prediction)
147  predictionResult = algorithm.compute()
148 
149 
150 def printResults():
151 
152  printNumericTables(
153  testGroundTruth,
154  predictionResult.get(classifier.prediction.prediction),
155  "Ground truth", "Classification results",
156  "Decision tree classification results (first 20 observations):",
157  20, flt64=False
158  )
159 
160 if __name__ == "__main__":
161 
162  trainModel()
163  testModel()
164  printResults()

For more complete information about compiler optimizations, see our Optimization Notice.