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

pca_transform_dense_batch.py

1 # file: pca_transform_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 #
43 # ! Content:
44 # ! Python example of PCA transformation algorithm.
45 # !*****************************************************************************
46 
47 #
48 
49 
50 #
51 
52 import os
53 import sys
54 import numpy as np
55 
56 import daal.algorithms.pca as pca
57 import daal.algorithms.pca.transform as pca_transform
58 from daal.data_management import DataSourceIface, FileDataSource
59 
60 utils_folder = os.path.realpath(os.path.abspath(os.path.dirname(os.path.dirname(os.path.dirname(__file__)))))
61 if utils_folder not in sys.path:
62  sys.path.insert(0, utils_folder)
63 from utils import printNumericTable
64 from daal.data_management import NumericTable
65 # Input data set parameters
66 datasetName = os.path.join('..', 'data', 'batch', 'pca_transform.csv')
67 
68 if __name__ == "__main__":
69 
70  # Retrieve the input data
71  dataSource = FileDataSource(datasetName,
72  DataSourceIface.doAllocateNumericTable,
73  DataSourceIface.doDictionaryFromContext)
74  dataSource.loadDataBlock()
75  data = dataSource.getNumericTable()
76 
77  # Create an algorithm
78  algorithm = pca.Batch(fptype=np.float64,method=pca.svdDense)
79 
80  # Set the algorithm input data
81  algorithm.input.setDataset(pca.data, data)
82 
83  # Set the algorithm normalization parameters (mean and variance)
84  # to be exported for transform and whitening parameter (eigenvalue)
85  # If whitening is not required eigenvalues should be removed
86  # The eigenvalues would be calculated in pca.eigenvalues table of result
87  # but would not be passed to dataForTranform collection
88  # algorithm.paramter.resultsToCompute = (pca.mean | pca.variance | pca.eigenvalue)
89 
90  algorithm.parameter.resultsToCompute = pca.mean | pca.variance | pca.eigenvalue;
91 
92  # Compute PCA
93  res = algorithm.compute()
94  # Output basis, eigenvalues and mean values
95  printNumericTable(res.get(pca.eigenvalues), "Eigenvalues:")
96  printNumericTable(res.get(pca.eigenvectors), "Eigenvectors:")
97 
98  eigenvaluesT = res.get(pca.eigenvalues)
99  printNumericTable(eigenvaluesT, "Eigenvalues kv:")
100 
101  meansT = res.get(pca.means)
102  printNumericTable(meansT, "Means kv:")
103 
104  #eigenvaluesT = res.getCollection(pca.eigenvalue)
105  variancesT = res.get(pca.variances)
106  printNumericTable(variancesT, "Variances kv:")
107 
108  # Create an algorithm
109  tralgorithm = pca_transform.Batch(fptype=np.float64)
110 
111  # Set lower and upper bounds for the algorithm
112  tralgorithm.parameter.nComponents = 2
113 
114  # Set an input object for the algorithm
115  tralgorithm.input.setTable(pca_transform.data, data)
116 
117  # Set an input object for the eigenvectors
118  tralgorithm.input.setTable(pca_transform.eigenvectors, res.get(pca.eigenvectors))
119 
120  # Set an input object for the eigenvectors
121  tralgorithm.input.setCollection(pca_transform.dataForTransform, res.getCollection(pca.dataForTransform))
122 
123  # Compute PCA transformation function
124  trres = tralgorithm.compute()
125 
126  printNumericTable(trres.get(pca.transform.transformedData), "Transformed data:");
127  #printNumericTable(data, "First rows of the input data:", 4)
128  #printNumericTable(trres.get(pca_transform.transformedData), "First rows of the min-max normalization result:", 4)

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