33 from daal
import step1Local, step2Master
34 from daal.algorithms.ridge_regression
import training, prediction
35 from daal.data_management
import DataSource, FileDataSource, NumericTable, HomogenNumericTable, MergedNumericTable
37 utils_folder = os.path.realpath(os.path.abspath(os.path.dirname(os.path.dirname(__file__))))
38 if utils_folder
not in sys.path:
39 sys.path.insert(0, utils_folder)
40 from utils
import printNumericTable
42 trainDatasetFileNames = [
43 os.path.join(
"..",
"data",
"distributed",
"linear_regression_train_1.csv"),
44 os.path.join(
"..",
"data",
"distributed",
"linear_regression_train_2.csv"),
45 os.path.join(
"..",
"data",
"distributed",
"linear_regression_train_3.csv"),
46 os.path.join(
"..",
"data",
"distributed",
"linear_regression_train_4.csv"),
50 testDatasetFileName = os.path.join(
"..",
"data",
"distributed",
"linear_regression_test.csv")
54 nDependentVariables = 2
59 masterAlgorithm = training.Distributed(step=step2Master)
61 for i
in range(nBlocks):
63 trainDataSource = FileDataSource(trainDatasetFileNames[i],
64 DataSource.notAllocateNumericTable,
65 DataSource.doDictionaryFromContext)
68 trainData = HomogenNumericTable(nFeatures, 0, NumericTable.doNotAllocate)
69 trainDependentVariables = HomogenNumericTable(nDependentVariables, 0, NumericTable.doNotAllocate)
70 mergedData = MergedNumericTable(trainData, trainDependentVariables)
73 trainDataSource.loadDataBlock(mergedData)
76 localAlgorithm = training.Distributed(step=step1Local)
79 localAlgorithm.input.set(training.data, trainData)
80 localAlgorithm.input.set(training.dependentVariables, trainDependentVariables)
83 presult = localAlgorithm.compute()
86 masterAlgorithm.input.add(training.partialModels, presult)
90 masterAlgorithm.compute()
93 trainingResult = masterAlgorithm.finalizeCompute()
95 printNumericTable(trainingResult.get(training.model).getBeta(),
"Ridge Regression coefficients:")
99 def testModel(trainingResult):
101 testDataSource = FileDataSource(testDatasetFileName,
102 DataSource.doAllocateNumericTable,
103 DataSource.doDictionaryFromContext)
106 testData = HomogenNumericTable(nFeatures, 0, NumericTable.doNotAllocate)
107 testGroundTruth = HomogenNumericTable(nDependentVariables, 0, NumericTable.doNotAllocate)
108 mergedData = MergedNumericTable(testData, testGroundTruth)
111 testDataSource.loadDataBlock(mergedData)
114 algorithm = prediction.Batch()
117 algorithm.input.setTable(prediction.data, testData)
118 algorithm.input.setModel(prediction.model, trainingResult.get(training.model))
121 predictionResult = algorithm.compute()
123 printNumericTable(predictionResult.get(prediction.prediction),
"Ridge Regression prediction results: (first 10 rows):", 10)
124 printNumericTable(testGroundTruth,
"Ground truth (first 10 rows):", 10)
127 if __name__ ==
"__main__":
128 trainingResult = trainModel()
129 testModel(trainingResult)