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

dt_reg_dense_batch.py

1 # file: dt_reg_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 
44 
45 import os
46 import sys
47 
48 from daal.algorithms.decision_tree.regression import prediction, training
49 from daal.data_management import (
50  FileDataSource, DataSourceIface, NumericTableIface, HomogenNumericTable, MergedNumericTable
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 printNumericTables
57 
58 DAAL_PREFIX = os.path.join('..', 'data')
59 
60 # Input data set parameters
61 trainDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'decision_tree_train.csv')
62 pruneDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'decision_tree_prune.csv')
63 testDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'decision_tree_test.csv')
64 
65 nFeatures = 5
66 
67 # Model object for the decision tree regression algorithm
68 model = None
69 predictionResult = None
70 testGroundTruth = None
71 
72 
73 def trainModel():
74  global model
75 
76  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
77  trainDataSource = FileDataSource(
78  trainDatasetFileName,
79  DataSourceIface.notAllocateNumericTable,
80  DataSourceIface.doDictionaryFromContext
81  )
82 
83  # Create Numeric Tables for training data and labels
84  trainData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
85  trainGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
86  mergedData = MergedNumericTable(trainData, trainGroundTruth)
87 
88  # Retrieve the data from the input file
89  trainDataSource.loadDataBlock(mergedData)
90 
91  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
92  pruneDataSource = FileDataSource(
93  pruneDatasetFileName,
94  DataSourceIface.notAllocateNumericTable,
95  DataSourceIface.doDictionaryFromContext
96  )
97 
98  # Create Numeric Tables for pruning data and labels
99  pruneData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
100  pruneGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
101  pruneMergedData = MergedNumericTable(pruneData, pruneGroundTruth)
102 
103  # Retrieve the data from the input file
104  pruneDataSource.loadDataBlock(pruneMergedData)
105 
106  # Create an algorithm object to train the decision tree regression model
107  algorithm = training.Batch()
108 
109  # Pass the training data set and dependent values to the algorithm
110  algorithm.input.set(training.data, trainData)
111  algorithm.input.set(training.dependentVariables, trainGroundTruth)
112  algorithm.input.set(training.dataForPruning, pruneData)
113  algorithm.input.set(training.dependentVariablesForPruning, pruneGroundTruth)
114 
115  # Train the decision tree regression model and retrieve the results of the training algorithm
116  trainingResult = algorithm.compute()
117  model = trainingResult.get(training.model)
118 
119 def testModel():
120  global testGroundTruth, predictionResult
121 
122  # Initialize FileDataSource<CSVFeatureManager> to retrieve the test data from a .csv file
123  testDataSource = FileDataSource(
124  testDatasetFileName,
125  DataSourceIface.notAllocateNumericTable,
126  DataSourceIface.doDictionaryFromContext
127  )
128 
129  # Create Numeric Tables for testing data and labels
130  testData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
131  testGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
132  mergedData = MergedNumericTable(testData, testGroundTruth)
133 
134  # Retrieve the data from input file
135  testDataSource.loadDataBlock(mergedData)
136 
137  # Create algorithm objects for decision tree regression prediction with the default method
138  algorithm = prediction.Batch()
139 
140  # Pass the testing data set and trained model to the algorithm
141  #print("Number of columns: {}".format(testData.getNumberOfColumns()))
142  algorithm.input.setTable(prediction.data, testData)
143  algorithm.input.setModel(prediction.model, model)
144 
145  # Compute prediction results and retrieve algorithm results
146  predictionResult = algorithm.compute()
147 
148 
149 def printResults():
150 
151  printNumericTables(testGroundTruth, predictionResult.get(prediction.prediction),
152  "Ground truth", "Regression results",
153  "Decision tree regression results (first 20 observations):",
154  20, flt64=False)
155 
156 if __name__ == "__main__":
157 
158  trainModel()
159  testModel()
160  printResults()

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