def testMerge(self):
if IGNORE_TEST:
return
NUM = 4
statistics = self.mkStatistics(NUM)
statistic = TimeseriesStatistic.merge(statistics)
statistic.calculate()
self.evaluateStatistic(statistic, count=NUM)
def merge(cls, bootstrapResults, fitter=None):
"""
Combines a list of BootstrapResult. Sets a fitter.
Parameter
---------
bootstrapResults: list-BootstrapResult
fitter: ModelFitter
Return
------
BootstrapResult
"""
if len(bootstrapResults) == 0:
raise ValueError("Must provide a non-empty list")
parameterDct = {p: [] for p in bootstrapResults[0].parameterDct}
numIteration = sum([r.numIteration for r in bootstrapResults])
bootstrapError = sum([b.bootstrapError for b in bootstrapResults])
fittedStatistic = None
if numIteration > 0:
# Merge the logs
logger = Logger.merge([b.logger for b in bootstrapResults])
# Merge the statistics for fitted timeseries
fittedStatistics = [
b.fittedStatistic for b in bootstrapResults
if b.fittedStatistic is not None
]
fittedStatistic = TimeseriesStatistic.merge(fittedStatistics)
# Accumulate the results
for bootstrapResult in bootstrapResults:
for parameter in parameterDct.keys():
parameterDct[parameter].extend(
bootstrapResult.parameterDct[parameter])
#
fitter.logger = Logger.merge([fitter.logger, logger])
bootstrapResult = BootstrapResult(fitter,
numIteration,
parameterDct,
fittedStatistic,
bootstrapError=bootstrapError)
bootstrapResult.setFitter(fitter)
return bootstrapResult