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

lin_reg_norm_eq_dense_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 
42 
43 
44 import os
45 import sys
46 
47 from daal import step1Local, step2Master
48 from daal.algorithms.linear_regression import training, prediction
49 from daal.data_management import (
50  DataSourceIface, FileDataSource, HomogenNumericTable, MergedNumericTable,NumericTableIface
51 )
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 printNumericTable
57 
58 DAAL_PREFIX = os.path.join('..', 'data')
59 
60 trainDatasetFileNames = [
61  os.path.join(DAAL_PREFIX, 'distributed', 'linear_regression_train_1.csv'),
62  os.path.join(DAAL_PREFIX, 'distributed', 'linear_regression_train_2.csv'),
63  os.path.join(DAAL_PREFIX, 'distributed', 'linear_regression_train_3.csv'),
64  os.path.join(DAAL_PREFIX, 'distributed', 'linear_regression_train_4.csv')
65 ]
66 
67 testDatasetFileName = os.path.join(DAAL_PREFIX, 'distributed', 'linear_regression_test.csv')
68 
69 nBlocks = 4
70 
71 nFeatures = 10 # Number of features in training and testing data sets
72 nDependentVariables = 2 # Number of dependent variables that correspond to each observation
73 
74 trainingResult = None
75 predictionResult = None
76 
77 
78 def trainModel():
79  global trainingResult
80 
81  # Create an algorithm object to build the final multiple linear regression model on the master node
82  masterAlgorithm = training.Distributed(step2Master)
83 
84  for i in range(nBlocks):
85  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
86  trainDataSource = FileDataSource(
87  trainDatasetFileNames[i], DataSourceIface.notAllocateNumericTable,
88  DataSourceIface.doDictionaryFromContext
89  )
90 
91  # Create Numeric Tables for training data and variables
92  trainData = HomogenNumericTable(nFeatures, 0, NumericTableIface.doNotAllocate)
93  trainDependentVariables = HomogenNumericTable(
94  nDependentVariables, 0, NumericTableIface.doNotAllocate
95  )
96  mergedData = MergedNumericTable(trainData, trainDependentVariables)
97 
98  # Retrieve the data from input file
99  trainDataSource.loadDataBlock(mergedData)
100 
101  # Create an algorithm object to train the multiple linear regression model based on the local-node data
102  localAlgorithm = training.Distributed(step1Local)
103 
104  # Pass a training data set and dependent values to the algorithm
105  localAlgorithm.input.set(training.data, trainData)
106  localAlgorithm.input.set(training.dependentVariables, trainDependentVariables)
107 
108  # Train the multiple linear regression model on the local-node data
109  # Set the local multiple linear regression model as input for the master-node algorithm
110  masterAlgorithm.input.add(training.partialModels, localAlgorithm.compute())
111 
112  # Merge and finalize the multiple linear regression model on the master node
113  masterAlgorithm.compute()
114 
115  # Retrieve the algorithm results
116  trainingResult = masterAlgorithm.finalizeCompute()
117  printNumericTable(trainingResult.get(training.model).getBeta(), "Linear Regression coefficients:")
118 
119 
120 def testModel():
121  global trainingResult, predictionResult
122 
123  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
124  testDataSource = FileDataSource(
125  testDatasetFileName, DataSourceIface.doAllocateNumericTable,
126  DataSourceIface.doDictionaryFromContext
127  )
128 
129  # Create Numeric Tables for testing data and ground truth values
130  testData = HomogenNumericTable(nFeatures, 0, NumericTableIface.doNotAllocate)
131  testGroundTruth = HomogenNumericTable(nDependentVariables, 0, NumericTableIface.doNotAllocate)
132  mergedData = MergedNumericTable(testData, testGroundTruth)
133 
134  # Retrieve the data from the input file
135  testDataSource.loadDataBlock(mergedData)
136 
137  # Create an algorithm object to predict values of multiple linear regression
138  algorithm = prediction.Batch()
139 
140  # Pass a testing data set and the trained model to the algorithm
141  algorithm.input.setTable(prediction.data, testData)
142  algorithm.input.setModel(prediction.model, trainingResult.get(training.model))
143 
144  # Predict values of multiple linear regression and retrieve the algorithm results
145  predictionResult = algorithm.compute()
146  printNumericTable(predictionResult.get(prediction.prediction), "Linear Regression prediction results: (first 10 rows):", 10)
147  printNumericTable(testGroundTruth, "Ground truth (first 10 rows):", 10)
148 
149 if __name__ == "__main__":
150 
151  trainModel()
152  testModel()

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