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

stump_dense_batch.py

1 # file: stump_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 
43 
44 
45 import os
46 import sys
47 
48 from daal.algorithms import classifier
49 from daal.algorithms.stump import training, prediction
50 from daal.data_management import (
51  FileDataSource, DataSourceIface, HomogenNumericTable, MergedNumericTable, NumericTableIface
52 )
53 
54 utils_folder = os.path.realpath(os.path.abspath(os.path.dirname(os.path.dirname(__file__))))
55 if utils_folder not in sys.path:
56  sys.path.insert(0, utils_folder)
57 from utils import printNumericTables
58 
59 DAAL_PREFIX = os.path.join('..', 'data')
60 
61 # Input data set parameters
62 nFeatures = 20
63 trainDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'stump_train.csv')
64 testDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'stump_test.csv')
65 
66 trainingResult = None
67 predictionResult = None
68 testGroundTruth = None
69 
70 
71 def trainModel():
72  global trainingResult
73  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
74  trainDataSource = FileDataSource(
75  trainDatasetFileName,
76  DataSourceIface.notAllocateNumericTable,
77  DataSourceIface.doDictionaryFromContext
78  )
79 
80  # Create Numeric Tables for training data and labels
81  trainData = HomogenNumericTable(nFeatures, 0, NumericTableIface.doNotAllocate)
82  trainGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.doNotAllocate)
83  mergedData = MergedNumericTable(trainData, trainGroundTruth)
84 
85  # Retrieve the data from the input file
86  trainDataSource.loadDataBlock(mergedData)
87 
88  # Create an algorithm object to train the stump model
89  algorithm = training.Batch()
90 
91  # Pass a training data set and dependent values to the algorithm
92  algorithm.input.set(classifier.training.data, trainData)
93  algorithm.input.set(classifier.training.labels, trainGroundTruth)
94 
95  # Compute and retrieve the algorithm results
96  trainingResult = algorithm.compute()
97 
98 
99 def testModel():
100  global predictionResult, testGroundTruth
101 
102  # Initialize FileDataSource<CSVFeatureManager> to retrieve the test data from a .csv file
103  testDataSource = FileDataSource(
104  testDatasetFileName,
105  DataSourceIface.notAllocateNumericTable,
106  DataSourceIface.doDictionaryFromContext
107  )
108 
109  # Create Numeric Tables for training data and labels
110  testData = HomogenNumericTable(nFeatures, 0, NumericTableIface.doNotAllocate)
111  testGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.doNotAllocate)
112  mergedData = MergedNumericTable(testData, testGroundTruth)
113 
114  # Retrieve the data from the input file
115  testDataSource.loadDataBlock(mergedData)
116 
117  # Create an algorithm object to train the stump model
118  algorithm = prediction.Batch()
119 
120  # Pass a training data set and dependent values to the algorithm
121  algorithm.input.setTable(classifier.prediction.data, testData)
122  algorithm.input.setModel(classifier.prediction.model,
123  trainingResult.get(classifier.training.model))
124 
125  # Compute and retrieve the algorithm Result class from classifier.prediction
126  predictionResult = algorithm.compute()
127 
128 
129 def printResults():
130  printNumericTables(
131  testGroundTruth,
132  predictionResult.get(classifier.prediction.prediction),
133  "Ground truth", "Classification results",
134  "Stump classification results (first 20 observations):", 20, flt64=False)
135 
136 if __name__ == "__main__":
137  trainModel()
138  testModel()
139  printResults()

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