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

df_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 import decision_forest
48 from daal.algorithms.decision_forest.regression import prediction, training
49 from daal.data_management import (
50  FileDataSource, DataSourceIface, NumericTableIface,
51  HomogenNumericTable, MergedNumericTable, data_feature_utils
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 trainDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'df_regression_train.csv')
63 testDatasetFileName = os.path.join(DAAL_PREFIX, 'batch', 'df_regression_test.csv')
64 
65 nFeatures = 13
66 
67 # Decision forest parameters
68 nTrees = 100
69 
70 # Model object for the decision forest regression algorithm
71 model = None
72 predictionResult = None
73 testGroundTruth = None
74 
75 
76 def trainModel():
77  global model
78 
79  # Initialize FileDataSource<CSVFeatureManager> to retrieve the input data from a .csv file
80  trainDataSource = FileDataSource(
81  trainDatasetFileName,
82  DataSourceIface.notAllocateNumericTable,
83  DataSourceIface.doDictionaryFromContext
84  )
85 
86  # Create Numeric Tables for training data and labels
87  trainData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
88  trainGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
89  mergedData = MergedNumericTable(trainData, trainGroundTruth)
90 
91  # Retrieve the data from the input file
92  trainDataSource.loadDataBlock(mergedData)
93 
94  # Get the dictionary and update it with additional information about data
95  dict = trainData.getDictionary()
96 
97  # Add a feature type to the dictionary
98  dict[3].featureType = data_feature_utils.DAAL_CATEGORICAL
99 
100  # Create an algorithm object to train the decision forest regression model
101  algorithm = training.Batch()
102  algorithm.parameter.nTrees = nTrees
103  algorithm.parameter.varImportance = decision_forest.training.MDA_Raw
104  algorithm.parameter.resultsToCompute = decision_forest.training.computeOutOfBagError
105 
106  # Pass the training data set and dependent values to the algorithm
107  algorithm.input.set(training.data, trainData)
108  algorithm.input.set(training.dependentVariable, trainGroundTruth)
109 
110  # Train the decision forest regression model and retrieve the results of the training algorithm
111  trainingResult = algorithm.compute()
112  model = trainingResult.get(training.model)
113  printNumericTable(trainingResult.getTable(training.variableImportance), "Variable importance results: ")
114  printNumericTable(trainingResult.getTable(training.outOfBagError), "OOB error: ")
115 
116 def testModel():
117  global testGroundTruth, predictionResult
118 
119  # Initialize FileDataSource<CSVFeatureManager> to retrieve the test data from a .csv file
120  testDataSource = FileDataSource(
121  testDatasetFileName,
122  DataSourceIface.notAllocateNumericTable,
123  DataSourceIface.doDictionaryFromContext
124  )
125 
126  # Create Numeric Tables for testing data and labels
127  testData = HomogenNumericTable(nFeatures, 0, NumericTableIface.notAllocate)
128  testGroundTruth = HomogenNumericTable(1, 0, NumericTableIface.notAllocate)
129  mergedData = MergedNumericTable(testData, testGroundTruth)
130 
131  # Retrieve the data from input file
132  testDataSource.loadDataBlock(mergedData)
133 
134  # Get the dictionary and update it with additional information about data
135  dict = testData.getDictionary()
136 
137  # Add a feature type to the dictionary
138  dict[3].featureType = data_feature_utils.DAAL_CATEGORICAL
139 
140  # Create algorithm objects for decision forest regression prediction with the default method
141  algorithm = prediction.Batch()
142 
143  # Pass the testing data set and trained model to the algorithm
144  algorithm.input.setTable(prediction.data, testData)
145  algorithm.input.set(prediction.model, model)
146 
147  # Compute prediction results and retrieve algorithm results
148  predictionResult = algorithm.compute()
149 
150 
151 def printResults():
152 
153  printNumericTable(
154  predictionResult.get(prediction.prediction),
155  "Decision forest prediction results (first 10 rows):", 10
156  )
157  printNumericTable(
158  testGroundTruth,
159  "Ground truth (first 10 rows):", 10
160  )
161 
162 if __name__ == "__main__":
163 
164  trainModel()
165  testModel()
166  printResults()

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