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

mn_naive_bayes_csr_distr.py

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

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