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

lin_reg_qr_dense_distr.py

1 # file: lin_reg_qr_dense_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.linear_regression import training, prediction
50 from daal.data_management import (
51  DataSourceIface, FileDataSource, 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 printNumericTable
58 
59 DAAL_PREFIX = os.path.join('..', 'data')
60 
61 # Input data set parameters
62 trainDatasetFileNames = [
63  os.path.join(DAAL_PREFIX, 'distributed', 'linear_regression_train_1.csv'),
64  os.path.join(DAAL_PREFIX, 'distributed', 'linear_regression_train_2.csv'),
65  os.path.join(DAAL_PREFIX, 'distributed', 'linear_regression_train_3.csv'),
66  os.path.join(DAAL_PREFIX, 'distributed', 'linear_regression_train_4.csv')
67 ]
68 
69 testDatasetFileName = os.path.join(DAAL_PREFIX, 'distributed', 'linear_regression_test.csv')
70 
71 nBlocks = 4
72 
73 nFeatures = 10 # Number of features in training and testing data sets
74 nDependentVariables = 2 # Number of dependent variables that correspond to each observation
75 
76 trainingResult = None
77 predictionResult = None
78 
79 
80 def trainModel():
81  global trainingResult
82 
83  # Create an algorithm object to build the final multiple linear regression model on the master node
84  masterAlgorithm = training.Distributed(step2Master, method=training.qrDense)
85 
86  for i in range(nBlocks):
87  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
88  trainDataSource = FileDataSource(
89  trainDatasetFileNames[i], DataSourceIface.notAllocateNumericTable,
90  DataSourceIface.doDictionaryFromContext
91  )
92 
93  # Create Numeric Tables for training data and dependent variables
94  trainData = HomogenNumericTable(nFeatures, 0, NumericTableIface.doNotAllocate)
95  trainDependentVariables = HomogenNumericTable(
96  nDependentVariables, 0, NumericTableIface.doNotAllocate
97  )
98  mergedData = MergedNumericTable(trainData, trainDependentVariables)
99 
100  # Retrieve the data from input file
101  trainDataSource.loadDataBlock(mergedData)
102 
103  # Create an algorithm object to train the multiple linear regression model based on the local-node data
104  localAlgorithm = training.Distributed(step1Local, method=training.qrDense)
105 
106  # Pass a training data set and dependent values to the algorithm
107  localAlgorithm.input.set(training.data, trainData)
108  localAlgorithm.input.set(training.dependentVariables, trainDependentVariables)
109 
110  # Train the multiple linear regression model on the local-node data
111  # Set the local multiple linear regression model as input for the master-node algorithm
112  masterAlgorithm.input.add(training.partialModels, localAlgorithm.compute())
113 
114  # Merge and finalize the multiple linear regression model on the master node
115  masterAlgorithm.compute()
116 
117  # Retrieve the algorithm results
118  trainingResult = masterAlgorithm.finalizeCompute()
119  printNumericTable(trainingResult.get(training.model).getBeta(), "Linear Regression coefficients:")
120 
121 
122 def testModel():
123 
124  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
125  testDataSource = FileDataSource(
126  testDatasetFileName, DataSourceIface.doAllocateNumericTable,
127  DataSourceIface.doDictionaryFromContext
128  )
129 
130  # Create Numeric Tables for testing data and ground truth values
131  testData = HomogenNumericTable(nFeatures, 0, NumericTableIface.doNotAllocate)
132  testGroundTruth = HomogenNumericTable(nDependentVariables, 0, NumericTableIface.doNotAllocate)
133  mergedData = MergedNumericTable(testData, testGroundTruth)
134 
135  # Retrieve the data from the input file
136  testDataSource.loadDataBlock(mergedData)
137 
138  # Create an algorithm object to predict values of multiple linear regression
139  algorithm = prediction.Batch()
140 
141  # Pass a testing data set and the trained model to the algorithm
142  algorithm.input.setTable(prediction.data, testData)
143  algorithm.input.setModel(prediction.model, trainingResult.get(training.model))
144 
145  # Predict values of multiple linear regression and retrieve the algorithm results
146  predictionResult = algorithm.compute()
147  printNumericTable(predictionResult.get(prediction.prediction), "Linear Regression prediction results: (first 10 rows):", 10)
148  printNumericTable(testGroundTruth, "Ground truth (first 10 rows):", 10)
149 
150 if __name__ == "__main__":
151 
152  trainModel()
153  testModel()

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