def backTest():
'''
回测
'''
# 读入贷款协议数据，交易流水数据和签约产品数据
reader = CMSBReader(cf.fieldName2fieldType)
trnFeatLoanFilenames = [os.path.join(cf.loanDir, month) for month in cf.trnFeatMonths]
trnFeatTransFilenames = [os.path.join(cf.transDir, month) for month in cf.trnFeatMonths]
trnFeatProdFilenames = [os.path.join(cf.prodDir, month) for month in cf.trnFeatMonths]
trnLabelLoanFilenames = [os.path.join(cf.loanDir, month) for month in cf.trnLabelMonths]
tstFeatLoanFilenames = [os.path.join(cf.loanDir, month) for month in cf.tstFeatMonths]
tstFeatTransFilenames = [os.path.join(cf.transDir, month) for month in cf.tstFeatMonths]
tstFeatProdFilenames = [os.path.join(cf.prodDir, month) for month in cf.tstFeatMonths]
tstLabelLoanFilenames = [os.path.join(cf.loanDir, month) for month in cf.tstLabelMonths]
loanFieldName2Index, trnFeatLoans, trnFeatCustNum2ProtolNums = reader.readLoans(trnFeatLoanFilenames)
transFieldName2Index, trnFeatTranss = reader.readTranss(trnFeatTransFilenames)
prodFieldName2Index, trnFeatProds = reader.readProds(trnFeatProdFilenames)
loanFieldName2Index, trnLabelLoans, trnLabelCustNum2ProtolNums = reader.readLoans(trnLabelLoanFilenames)
loanFieldName2Index, tstFeatLoans, tstFeatCustNum2ProtolNums = reader.readLoans(tstFeatLoanFilenames)
transFieldName2Index, tstFeatTranss = reader.readTranss(tstFeatTransFilenames)
prodFieldName2Index, tstFeatProds = reader.readProds(tstFeatProdFilenames)
loanFieldName2Index, tstLabelLoans, tstLabelCustNum2ProtolNums = reader.readLoans(tstLabelLoanFilenames)
# 过滤贷款协议数据
trnFeatLoans = filterLoans(cf.filterNames, loanFieldName2Index, trnFeatLoans, trnFeatCustNum2ProtolNums)
tstFeatLoans = filterLoans(cf.filterNames, loanFieldName2Index, tstFeatLoans, tstFeatCustNum2ProtolNums)
# ----------------------------------------------------------------------
# 生成样本
trn_samples_builder = SamplesBuilder(loanFieldName2Index, trnFeatLoans, trnFeatCustNum2ProtolNums, trnLabelLoans,
transFieldName2Index, trnFeatTranss, prodFieldName2Index, trnFeatProds)
tst_samples_builder = SamplesBuilder(loanFieldName2Index, tstFeatLoans, tstFeatCustNum2ProtolNums, tstLabelLoans,
transFieldName2Index, tstFeatTranss, prodFieldName2Index, tstFeatProds)
trn_samples = trn_samples_builder.buildSample()
tst_samples = tst_samples_builder.buildSample()
# ----------------------------------------------------------------------
# 训练模型并预测
fit_predict(cf.modelName, cf.modelParam, trn_samples, tst_samples)
# 保存样本
trn_samples.save(cf.trnSampFilename)
tst_samples.save(cf.tstSampFilename)
# 分析样本
analyze_samples(trn_samples, tst_samples, 20, cf.analyFilename)
# 计算评价指标
gen_metrics(tst_samples, cf.metricFilename)
from OLP.Readers.CMSBReaders import CMSBReader import config from OLP.Readers.ReaderTools import UniPrinter from OLP.Readers.FeatureBuilder import FeatureBuilder from OLP.Readers.TransCounter import TransCounter from OLP.Readers.ProdContactCounter import ProdContactCounter from OLP.Readers.LoanCounter import LoanCounter #tableObject = CMSBReader(config.fieldName2fieldType) #table = tableObject.readProds(['prods.txt',]) #UniPrinter().pprint(table) #tableFiltered = ProdContactCounter(table, config.trnFeatMonths[-1]+'/31').countProdContact() #UniPrinter().pprint(tableFiltered) tableObject = CMSBReader(config.fieldName2fieldType) loanTable = tableObject.readLoans(['loan.txt','loan2.txt']) loanCounter = LoanCounter(loanTable).countLoan() transTable = tableObject.readTranss(['trans.txt',]) transCounter = TransCounter(transTable).countProp() prodTable = tableObject.readProds(['prods.txt',]) prodCounter = ProdContactCounter(prodTable, config.trnFeatMonths[-1]+'/31').countProdContact() UniPrinter().pprint(loanTable) UniPrinter().pprint(transTable) UniPrinter().pprint(prodTable) print 'haha' UniPrinter().pprint(loanCounter) UniPrinter().pprint(transCounter) UniPrinter().pprint(prodCounter)
return filter_
if __name__ == '__main__':
'''
main函数为测试代码，实际运行中不需要。
'''
from OLP.Readers.CMSBReaders import CMSBReader
import config
filenames = [
'2014-02-28LoanInformation.txt', '2014-03-31LoanInformation.txt',
'2014-04-30LoanInformation.txt', '2014-05-31LoanInformation.txt'
]
s = CMSBReader(config.fieldName2fieldType)
title2index, loans, custo2protol = s.readLoans(filenames)
print(len(loans))
print(custo2protol)
filter1 = CleanedLoanFilter()
filter2 = CustCodeFilter()
filter3 = ThisMonthLoanFilter()
title2index, loans, custo2protol = filter1.filter(title2index, loans,
custo2protol)
title2index, loans, custo2protol = filter2.filter(title2index, loans,
custo2protol)
title2index, loans, custo2protol = filter3.filter(title2index, loans,
custo2protol)
mod = __import__(modName, globals(), locals(), [clsName], -1)
cls = getattr(mod, clsName)
filter_ = cls(**param)
return filter_
if __name__ == '__main__':
'''
main函数为测试代码，实际运行中不需要。
'''
from OLP.Readers.CMSBReaders import CMSBReader
import config
filenames = ['2014-02-28LoanInformation.txt', '2014-03-31LoanInformation.txt', '2014-04-30LoanInformation.txt', '2014-05-31LoanInformation.txt']
s = CMSBReader(config.fieldName2fieldType)
title2index, loans, custo2protol = s.readLoans(filenames)
print(len(loans))
print(custo2protol)
filter1 = CleanedLoanFilter()
filter2 = CustCodeFilter()
filter3 = ThisMonthLoanFilter()
title2index, loans, custo2protol = filter1.filter(title2index, loans, custo2protol)
title2index, loans, custo2protol = filter2.filter(title2index, loans, custo2protol)
title2index, loans, custo2protol = filter3.filter(title2index, loans, custo2protol)
#print loans
print(len(loans))
print(custo2protol)
from OLP.Readers.CMSBReaders import CMSBReader
import config
from OLP.Readers.ReaderTools import UniPrinter
from OLP.Readers.FeatureBuilder import FeatureBuilder
from OLP.Readers.TransCounter import TransCounter
from OLP.Readers.ProdContactCounter import ProdContactCounter
from OLP.Readers.LoanCounter import LoanCounter
#tableObject = CMSBReader(config.fieldName2fieldType)
#table = tableObject.readProds(['prods.txt',])
#UniPrinter().pprint(table)
#tableFiltered = ProdContactCounter(table, config.trnFeatMonths[-1]+'/31').countProdContact()
#UniPrinter().pprint(tableFiltered)
tableObject = CMSBReader(config.fieldName2fieldType)
loanTable = tableObject.readLoans(['loan.txt', 'loan2.txt'])
loanCounter = LoanCounter(loanTable).countLoan()
transTable = tableObject.readTranss([
'trans.txt',
])
transCounter = TransCounter(transTable).countProp()
prodTable = tableObject.readProds([
'prods.txt',
])
prodCounter = ProdContactCounter(prodTable, config.trnFeatMonths[-1] +
'/31').countProdContact()
UniPrinter().pprint(loanTable)
UniPrinter().pprint(transTable)