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

df_reg_dense_batch.py

1 # file: df_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 import decision_forest
49 from daal.algorithms.decision_forest.regression import prediction, training
50 from daal.data_management import (
51  FileDataSource, DataSourceIface, NumericTableIface,
52  HomogenNumericTable, MergedNumericTable, data_feature_utils
53 )
54 
55 utils_folder = os.path.realpath(os.path.abspath(os.path.dirname(os.path.dirname(__file__))))
56 if utils_folder not in sys.path:
57  sys.path.insert(0, utils_folder)
58 from utils import printNumericTable
59 
60 DAAL_PREFIX = os.path.join('..', 'data')
61 
62 # Input data set parameters
63 trainDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'df_regression_train.csv')
64 testDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'df_regression_test.csv')
65 
66 nFeatures = 13
67 
68 # Decision forest parameters
69 nTrees = 100
70 
71 # Model object for the decision forest regression algorithm
72 model = None
73 predictionResult = None
74 testGroundTruth = None
75 
76 
77 def trainModel():
78  global model
79 
80  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
81  trainDataSource = FileDataSource(
82  trainDatasetFileName,
83  DataSourceIface.notAllocateNumericTable,
84  DataSourceIface.doDictionaryFromContext
85  )
86 
87  # Create Numeric Tables for training data and labels
88  trainData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
89  trainGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
90  mergedData = MergedNumericTable(trainData, trainGroundTruth)
91 
92  # Retrieve the data from the input file
93  trainDataSource.loadDataBlock(mergedData)
94 
95  # Get the dictionary and update it with additional information about data
96  dict = trainData.getDictionary()
97 
98  # Add a feature type to the dictionary
99  dict[3].featureType = data_feature_utils.DAAL_CATEGORICAL
100 
101  # Create an algorithm object to train the decision forest regression model
102  algorithm = training.Batch()
103  algorithm.parameter.nTrees = nTrees
104  algorithm.parameter.varImportance = decision_forest.training.MDA_Raw
105  algorithm.parameter.resultsToCompute = decision_forest.training.computeOutOfBagError|decision_forest.training.computeOutOfBagErrorPerObservation;
106 
107  # Pass the training data set and dependent values to the algorithm
108  algorithm.input.set(training.data, trainData)
109  algorithm.input.set(training.dependentVariable, trainGroundTruth)
110 
111  # Train the decision forest regression model and retrieve the results of the training algorithm
112  trainingResult = algorithm.compute()
113  model = trainingResult.get(training.model)
114  printNumericTable(trainingResult.getTable(training.variableImportance), "Variable importance results: ")
115  printNumericTable(trainingResult.getTable(training.outOfBagError), "OOB error: ")
116  printNumericTable(trainingResult.getTable(training.outOfBagError), "OOB error (first 10 rows): ", 10)
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  # Get the dictionary and update it with additional information about data
137  dict = testData.getDictionary()
138 
139  # Add a feature type to the dictionary
140  dict[3].featureType = data_feature_utils.DAAL_CATEGORICAL
141 
142  # Create algorithm objects for decision forest regression prediction with the default method
143  algorithm = prediction.Batch()
144 
145  # Pass the testing data set and trained model to the algorithm
146  algorithm.input.setTable(prediction.data, testData)
147  algorithm.input.set(prediction.model, model)
148 
149  # Compute prediction results and retrieve algorithm results
150  predictionResult = algorithm.compute()
151 
152 
153 def printResults():
154 
155  printNumericTable(
156  predictionResult.get(prediction.prediction),
157  "Decision forest prediction results (first 10 rows):", 10
158  )
159  printNumericTable(
160  testGroundTruth,
161  "Ground truth (first 10 rows):", 10
162  )
163 
164 if __name__ == "__main__":
165 
166  trainModel()
167  testModel()
168  printResults()

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