def createLeaveOneOutJobs(
srcRoot,
jobDir,
jobNameFmt,
finalJobName,
outputDir,
numObservations,
dataLoader,
lambdas,
shouldRegBias,
matlabDeclarations,
):
jobNamePart = jobNameFmt.format("NA", "NA")
foldIndices = range(1, numObservations + 1)
if matlabDeclarations is None:
matlabDeclarations = list()
logRegRoot = os.path.join(srcRoot, "shared", "logisticRegression")
addpathFmt = "addpath(genpath('{0}'));"
matlabDeclarations.append(addpathFmt.format(logRegRoot))
if shouldRegBias:
shouldRegBiasStr = "true"
else:
shouldRegBiasStr = "false"
cvFuncFmt = "LogRegSingleLambdaSingleFoldCVFromLoader({0}, {1}, {2}, {3}, {4})".format(
dataLoader, "{0}", numObservations, "{1}", shouldRegBiasStr
)
looJobDictionary = QSubHelpers.createGridJobs(
jobDir, outputDir, jobNameFmt, cvFuncFmt, matlabDeclarations, [lambdas, foldIndices], 50
)
mergeJobDictionary = QSubHelpers.createMergeAllJobOutputsJob(
jobDir, outputDir, outputDir, jobNameFmt, "loo_mrg_" + jobNamePart, [lambdas, foldIndices]
)
finalizeFunc = "LogRegChooseLambdaAndTrainAll({0}, {1}, '{2}')".format(
dataLoader, shouldRegBiasStr, os.path.join(outputDir, "output_loo_mrg_" + jobNamePart + ".mat")
)
finalizeJobDictionary = QSubHelpers.createGridJobs(
jobDir, outputDir, finalJobName, finalizeFunc, matlabDeclarations, ()
)
finalJobManifestName = "looLogRegFinalJobManifest_" + jobNamePart + ".txt"
return [looJobDictionary, mergeJobDictionary, finalizeJobDictionary]
def writeDataSizes(srcRoot, jobDir, outputDir, jobNameSuffixFmt, dataLoaderFmt, matlabDeclarations, parameters):
if matlabDeclarations is None:
matlabDeclarations = list()
logRegRoot = os.path.join(srcRoot, "shared", "logisticRegression")
addpathFmt = "addpath(genpath('{0}'));"
matlabDeclarations.append(addpathFmt.format(logRegRoot))
jobNameFmt = "plllogreg_getsz_" + jobNameSuffixFmt
writeSizeFuncFmt = "WriteSize( {0}, '{1}' );".format(
dataLoaderFmt, os.path.join(outputDir, "output_" + jobNameFmt + ".txt")
)
writeDataSizeJobDict = QSubHelpers.createGridJobs(
jobDir, outputDir, jobNameFmt, writeSizeFuncFmt, matlabDeclarations, parameters
)
QSubHelpers.runScripts(jobDir, outputDir, writeDataSizeJobDict, 100, None, None)
return os.path.join(outputDir, "output_" + jobNameFmt + ".txt")
