Python* API Reference for Intel® Data Analytics Acceleration Library 2018 Update 2

dt_cls_dense_batch.py

1 # file: dt_cls_dense_batch.py
2 #===============================================================================
3 # Copyright 2014-2018 Intel Corporation
4 # All Rights Reserved.
5 #
6 # If this software was obtained under the Intel Simplified Software License,
7 # the following terms apply:
8 #
9 # The source code, information and material ("Material") contained herein is
10 # owned by Intel Corporation or its suppliers or licensors, and title to such
11 # Material remains with Intel Corporation or its suppliers or licensors. The
12 # Material contains proprietary information of Intel or its suppliers and
13 # licensors. The Material is protected by worldwide copyright laws and treaty
14 # provisions. No part of the Material may be used, copied, reproduced,
15 # modified, published, uploaded, posted, transmitted, distributed or disclosed
16 # in any way without Intel's prior express written permission. No license under
17 # any patent, copyright or other intellectual property rights in the Material
18 # is granted to or conferred upon you, either expressly, by implication,
19 # inducement, estoppel or otherwise. Any license under such intellectual
20 # property rights must be express and approved by Intel in writing.
21 #
22 # Unless otherwise agreed by Intel in writing, you may not remove or alter this
23 # notice or any other notice embedded in Materials by Intel or Intel's
24 # suppliers or licensors in any way.
25 #
26 #
27 # If this software was obtained under the Apache License, Version 2.0 (the
28 # "License"), the following terms apply:
29 #
30 # You may not use this file except in compliance with the License. You may
31 # obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0
32 #
33 #
34 # Unless required by applicable law or agreed to in writing, software
35 # distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
36 # WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
37 #
38 # See the License for the specific language governing permissions and
39 # limitations under the License.
40 #===============================================================================
41 
42 
44 
45 import os
46 import sys
47 
48 from daal.algorithms.decision_tree.classification import prediction, training
49 from daal.algorithms import classifier
50 from daal.data_management import (
51  FileDataSource, DataSourceIface, NumericTableIface, HomogenNumericTable, MergedNumericTable
52 )
53 utils_folder = os.path.realpath(os.path.abspath(os.path.dirname(os.path.dirname(__file__))))
54 if utils_folder not in sys.path:
55  sys.path.insert(0, utils_folder)
56 from utils import printNumericTables
57 
58 DAAL_PREFIX = os.path.join('..', 'data')
59 
60 # Input data set parameters
61 trainDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'decision_tree_train.csv')
62 pruneDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'decision_tree_prune.csv')
63 testDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'decision_tree_test.csv')
64 
65 nFeatures = 5
66 nClasses = 5
67 
68 # Model object for the decision tree classification algorithm
69 model = None
70 predictionResult = None
71 testGroundTruth = None
72 
73 
74 def trainModel():
75  global model
76 
77  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
78  trainDataSource = FileDataSource(
79  trainDatasetFileName,
80  DataSourceIface.notAllocateNumericTable,
81  DataSourceIface.doDictionaryFromContext
82  )
83 
84  # Create Numeric Tables for training data and labels
85  trainData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
86  trainGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
87  mergedData = MergedNumericTable(trainData, trainGroundTruth)
88 
89  # Retrieve the data from the input file
90  trainDataSource.loadDataBlock(mergedData)
91 
92  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
93  pruneDataSource = FileDataSource(
94  pruneDatasetFileName,
95  DataSourceIface.notAllocateNumericTable,
96  DataSourceIface.doDictionaryFromContext
97  )
98 
99  # Create Numeric Tables for pruning data and labels
100  pruneData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
101  pruneGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
102  pruneMergedData = MergedNumericTable(pruneData, pruneGroundTruth)
103 
104  # Retrieve the data from the input file
105  pruneDataSource.loadDataBlock(pruneMergedData)
106 
107  # Create an algorithm object to train the decision tree classification model
108  algorithm = training.Batch(nClasses)
109 
110  # Pass the training data set and dependent values to the algorithm
111  algorithm.input.set(classifier.training.data, trainData)
112  algorithm.input.set(classifier.training.labels, trainGroundTruth)
113  algorithm.input.setTable(training.dataForPruning, pruneData)
114  algorithm.input.setTable(training.labelsForPruning, pruneGroundTruth)
115 
116  # Train the decision tree classification model and retrieve the results of the training algorithm
117  trainingResult = algorithm.compute()
118  model = trainingResult.get(classifier.training.model)
119 
120 def testModel():
121  global testGroundTruth, predictionResult
122 
123  # Initialize FileDataSource<CSVFeatureManager> to retrieve the test data from a .csv file
124  testDataSource = FileDataSource(
125  testDatasetFileName,
126  DataSourceIface.notAllocateNumericTable,
127  DataSourceIface.doDictionaryFromContext
128  )
129 
130  # Create Numeric Tables for testing data and labels
131  testData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
132  testGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
133  mergedData = MergedNumericTable(testData, testGroundTruth)
134 
135  # Retrieve the data from input file
136  testDataSource.loadDataBlock(mergedData)
137 
138  # Create algorithm objects for decision tree classification prediction with the default method
139  algorithm = prediction.Batch()
140 
141  # Pass the testing data set and trained model to the algorithm
142  #print("Number of columns: {}".format(testData.getNumberOfColumns()))
143  algorithm.input.setTable(classifier.prediction.data, testData)
144  algorithm.input.setModel(classifier.prediction.model, model)
145 
146  # Compute prediction results and retrieve algorithm results
147  # (Result class from classifier.prediction)
148  predictionResult = algorithm.compute()
149 
150 
151 def printResults():
152 
153  printNumericTables(
154  testGroundTruth,
155  predictionResult.get(classifier.prediction.prediction),
156  "Ground truth", "Classification results",
157  "Decision tree classification results (first 20 observations):",
158  20, flt64=False
159  )
160 
161 if __name__ == "__main__":
162 
163  trainModel()
164  testModel()
165  printResults()

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