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

dt_reg_dense_batch.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.algorithms.decision_tree.regression import prediction, training
48 from daal.data_management import (
49  FileDataSource, DataSourceIface, NumericTableIface, HomogenNumericTable, MergedNumericTable
50 )
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
56 
57 DAAL_PREFIX = os.path.join('..', 'data')
58 
59 # Input data set parameters
60 trainDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'decision_tree_train.csv')
61 pruneDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'decision_tree_prune.csv')
62 testDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'decision_tree_test.csv')
63 
64 nFeatures = 5
65 
66 # Model object for the decision tree regression algorithm
67 model = None
68 predictionResult = None
69 testGroundTruth = None
70 
71 
72 def trainModel():
73  global model
74 
75  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
76  trainDataSource = FileDataSource(
77  trainDatasetFileName,
78  DataSourceIface.notAllocateNumericTable,
79  DataSourceIface.doDictionaryFromContext
80  )
81 
82  # Create Numeric Tables for training data and labels
83  trainData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
84  trainGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
85  mergedData = MergedNumericTable(trainData, trainGroundTruth)
86 
87  # Retrieve the data from the input file
88  trainDataSource.loadDataBlock(mergedData)
89 
90  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
91  pruneDataSource = FileDataSource(
92  pruneDatasetFileName,
93  DataSourceIface.notAllocateNumericTable,
94  DataSourceIface.doDictionaryFromContext
95  )
96 
97  # Create Numeric Tables for pruning data and labels
98  pruneData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
99  pruneGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
100  pruneMergedData = MergedNumericTable(pruneData, pruneGroundTruth)
101 
102  # Retrieve the data from the input file
103  pruneDataSource.loadDataBlock(pruneMergedData)
104 
105  # Create an algorithm object to train the decision tree regression model
106  algorithm = training.Batch()
107 
108  # Pass the training data set and dependent values to the algorithm
109  algorithm.input.set(training.data, trainData)
110  algorithm.input.set(training.dependentVariables, trainGroundTruth)
111  algorithm.input.set(training.dataForPruning, pruneData)
112  algorithm.input.set(training.dependentVariablesForPruning, pruneGroundTruth)
113 
114  # Train the decision tree regression model and retrieve the results of the training algorithm
115  trainingResult = algorithm.compute()
116  model = trainingResult.get(training.model)
117 
118 def testModel():
119  global testGroundTruth, predictionResult
120 
121  # Initialize FileDataSource<CSVFeatureManager> to retrieve the test data from a .csv file
122  testDataSource = FileDataSource(
123  testDatasetFileName,
124  DataSourceIface.notAllocateNumericTable,
125  DataSourceIface.doDictionaryFromContext
126  )
127 
128  # Create Numeric Tables for testing data and labels
129  testData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
130  testGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
131  mergedData = MergedNumericTable(testData, testGroundTruth)
132 
133  # Retrieve the data from input file
134  testDataSource.loadDataBlock(mergedData)
135 
136  # Create algorithm objects for decision tree regression prediction with the default method
137  algorithm = prediction.Batch()
138 
139  # Pass the testing data set and trained model to the algorithm
140  print("Number of columns: {}".format(testData.getNumberOfColumns()))
141  algorithm.input.setTable(prediction.data, testData)
142  algorithm.input.setModel(prediction.model, model)
143 
144  # Compute prediction results and retrieve algorithm results
145  predictionResult = algorithm.compute()
146 
147 
148 def printResults():
149 
150  printNumericTables(testGroundTruth, predictionResult.get(prediction.prediction),
151  "Ground truth", "Regression results",
152  "Decision tree regression results (first 20 observations):",
153  20, flt64=False)
154 
155 if __name__ == "__main__":
156 
157  trainModel()
158  testModel()
159  printResults()

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