import ensembles.ensemble as evaluation
import os, glob, csv
import utility.dataoperations as dataoperations
from utility.enums import DataGenerationPattern, Processor, RnnType
import utility.run as run
import ensembles.pruning as pruningWrapper
import datadefinitions.cargo2000 as cargo2000
train_args = {
'datageneration_pattern': DataGenerationPattern.Fit,
'datadefinition': cargo2000.Cargo2000(),
'processor': Processor.GPU,
'cudnn': True,
#'bagging': True,
#'bagging_size': bagging_size,
'validationdata_split': 0.05,
'testdata_split': 0.3333,
'max_sequencelength': 50,
'batch_size': 64,
'neurons': 100,
'dropout': 0,
#'max_epochs': max_epochs,
'layers': 2,
#'save_model': False,
#'adaboost': None
}
args = run.Do_Preprocessing(**train_args)
args['test_sentences'] = dataoperations.CreateSentences(args['testdata'])
test_x, test_y = evaluation.prepare_data(args, 'testdata')
import utility.run
import os
import sys
from utility.enums import DataGenerationPattern, Processor, RnnType
# datasets to test
import datadefinitions.cargo2000 as cargo2000
import datadefinitions.cargo2000generic as cargo2000generic
import datadefinitions.bpi2012 as bpi2012
import datadefinitions.bpi2017 as bpi2017
import datadefinitions.bpi2018 as bpi2018
# test all testsets
for i, test_set in enumerate([
cargo2000.Cargo2000(),
cargo2000generic.Cargo2000(),
bpi2012.BPI2012(),
bpi2017.BPI2017(),
bpi2018.BPI2018()
]):
# run with low profile and default values to speed up tests
utility.run.Train_And_Evaluate(
datageneration_pattern=DataGenerationPattern.Generator,
datadefinition=test_set,
running=i,
bagging=True,
bagging_size=0.05,
validationdata_split=0.05,
testdata_split=0.05,
max_sequencelength=50,
batch_size=64,
import utility.run
import os
import sys
from utility.enums import DataGenerationPattern, Processor, RnnType
# datasets to test
import datadefinitions.cargo2000 as cargo2000
import datadefinitions.cargo2000generic as cargo2000generic
import datadefinitions.bpi2012 as bpi2012
import datadefinitions.bpi2017 as bpi2017
import datadefinitions.bpi2018 as bpi2018
import datadefinitions.road_traffic_fine_management as road_traffic_fine_management
import datadefinitions.sepsis as sepsis
# test all testsets
for i, test_set in enumerate([cargo2000.Cargo2000(),
cargo2000generic.Cargo2000(),
bpi2012.BPI2012(),
bpi2017.BPI2017(),
bpi2018.BPI2018(),
road_traffic_fine_management.RoadTrafficFine(),
sepsis.Sepsis()]):
# run with low profile and default values to speed up tests
utility.run.Train_And_Evaluate(
datageneration_pattern = DataGenerationPattern.Generator,
datadefinition=test_set,
running=i,
bagging=True,
bagging_size=0.05,
validationdata_split = 0.05,
testdata_split = 0.05,
import uuid
from utility.enums import DataGenerationPattern, Processor, RnnType
""" # check for env variable for cpu/gpu environment detection (CONDUCTOR)
TODO: implement
processorType = os.environ.get("CONDUCTOR_TYPE")
if processorType == "cpu":
print("cpu environment detected")
elif processorType == "gpu":
print("gpu environment detected")
else:
print("unknown environment detected, defaulting to gpu")
processorType = Processor.CPU """
# import data definition (dataset definitions in /datadefinitions/ folder)
import datadefinitions.cargo2000 as datadef
datadef = datadef.Cargo2000()
# cmd parameters
if len(sys.argv) > 1:
param = float(sys.argv[1])
# generate guid
guid = uuid.uuid4()
# call
utility.run.Train_And_Evaluate(
#data
eventlog="datasets/cargo2000.csv", # file to read in
datadefinition=datadef, # the data / matrix definitions
running=guid, # iterable / suffix
datageneration_pattern=DataGenerationPattern.