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

mn_naive_bayes_csr_distr.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 ## <a name="DAAL-EXAMPLE-PY-MULTINOMIAL_NAIVE_BAYES_CSR_DISTRIBUTED"></a>
42 ## \example mn_naive_bayes_csr_distr.py
43 
44 import os
45 import sys
46 
47 from daal import step1Local, step2Master
48 from daal.algorithms import classifier
49 from daal.algorithms.multinomial_naive_bayes import training, prediction
50 from daal.data_management import FileDataSource, DataSourceIface
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, createSparseTable
56 
57 DAAL_PREFIX = os.path.join('..', 'data')
58 
59 # Input data set parameters
60 trainDatasetFileNames = [
61  os.path.join(DAAL_PREFIX, 'batch', 'naivebayes_train_csr.csv'),
62  os.path.join(DAAL_PREFIX, 'batch', 'naivebayes_train_csr.csv'),
63  os.path.join(DAAL_PREFIX, 'batch', 'naivebayes_train_csr.csv'),
64  os.path.join(DAAL_PREFIX, 'batch', 'naivebayes_train_csr.csv')
65 ]
66 
67 trainGroundTruthFileNames = [
68  os.path.join(DAAL_PREFIX, 'batch', 'naivebayes_train_labels.csv'),
69  os.path.join(DAAL_PREFIX, 'batch', 'naivebayes_train_labels.csv'),
70  os.path.join(DAAL_PREFIX, 'batch', 'naivebayes_train_labels.csv'),
71  os.path.join(DAAL_PREFIX, 'batch', 'naivebayes_train_labels.csv')
72 ]
73 
74 testDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'naivebayes_test_csr.csv')
75 testGroundTruthFileName = os.path.join(DAAL_PREFIX, 'batch', 'naivebayes_test_labels.csv')
76 
77 nClasses = 20
78 nBlocks = 4
79 nTrainVectorsInBlock = 8000
80 nTestObservations = 2000
81 
82 trainingResult = None
83 predictionResult = None
84 trainData = [0] * nBlocks
85 testData = None
86 
87 
88 def trainModel():
89  global trainData, trainingResult
90 
91  masterAlgorithm = training.Distributed(step2Master, nClasses, method=training.fastCSR)
92 
93  for i in range(nBlocks):
94  # Read trainDatasetFileNames and create a numeric table to store the input data
95  trainData[i] = createSparseTable(trainDatasetFileNames[i])
96 
97  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
98  trainLabelsSource = FileDataSource(
99  trainGroundTruthFileNames[i], DataSourceIface.doAllocateNumericTable,
100  DataSourceIface.doDictionaryFromContext
101  )
102 
103  # Retrieve the data from an input file
104  trainLabelsSource.loadDataBlock(nTrainVectorsInBlock)
105 
106  # Create an algorithm object to train the Naive Bayes model on the local-node data
107  localAlgorithm = training.Distributed(step1Local, nClasses, method=training.fastCSR)
108 
109  # Pass a training data set and dependent values to the algorithm
110  localAlgorithm.input.set(classifier.training.data, trainData[i])
111  localAlgorithm.input.set(classifier.training.labels, trainLabelsSource.getNumericTable())
112 
113  # Build the Naive Bayes model on the local node
114  # Set the local Naive Bayes model as input for the master-node algorithm
115  masterAlgorithm.input.add(training.partialModels, localAlgorithm.compute())
116 
117  # Merge and finalize the Naive Bayes model on the master node
118  masterAlgorithm.compute()
119  trainingResult = masterAlgorithm.finalizeCompute() # Retrieve the algorithm results
120 
121 
122 def testModel():
123  global predictionResult, testData
124 
125  # Read testDatasetFileName and create a numeric table to store the input data
126  testData = createSparseTable(testDatasetFileName)
127 
128  # Create an algorithm object to predict Naive Bayes values
129  algorithm = prediction.Batch(nClasses, method=prediction.fastCSR)
130 
131  # Pass a testing data set and the trained model to the algorithm
132  algorithm.input.setTable(classifier.prediction.data, testData)
133  algorithm.input.setModel(classifier.prediction.model, trainingResult.get(classifier.training.model))
134 
135  # Predict Naive Bayes values (Result class from classifier.prediction)
136  predictionResult = algorithm.compute() # Retrieve the algorithm results
137 
138 
139 def printResults():
140 
141  testGroundTruth = FileDataSource(
142  testGroundTruthFileName, DataSourceIface.doAllocateNumericTable,
143  DataSourceIface.doDictionaryFromContext
144  )
145  testGroundTruth.loadDataBlock(nTestObservations)
146 
147  printNumericTables(
148  testGroundTruth.getNumericTable(),
149  predictionResult.get(classifier.prediction.prediction),
150  "Ground truth", "Classification results",
151  "NaiveBayes classification results (first 20 observations):", 20, 15, flt64=False
152  )
153 
154 if __name__ == "__main__":
155 
156  trainModel()
157  testModel()
158  printResults()

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