def load_scenarios(
self,
fps_d=None, #container of filepaths
base_dir=None, #base directory to add
):
#=======================================================================
# defaults
#=======================================================================
log = self.logger.getChild('load_scenarios')
if fps_d is None: fps_d = self.fps_d
if base_dir is None: base_dir = self.base_dir
log.info('on %i scenarios' % len(fps_d))
#=======================================================================
# precheck
#=======================================================================
assert len(fps_d) >= 2
assert isinstance(fps_d, dict)
assert len(fps_d.values()) == len(set(
fps_d.values())), 'must specify unique Control filepaths'
#=======================================================================
# relatives
#=======================================================================
if not self.absolute_fp:
fps_d = {k: os.path.join(base_dir, v) for k, v in fps_d.items()}
for k, fp in fps_d.items():
assert isinstance(fp, str)
assert os.path.exists(fp), 'bad filepath: %s' % (fp)
#=======================================================================
# build each scenario----
#=======================================================================
"""needs to be a strong reference or the workers die!"""
d = dict() #start a weak reference container
for i, (tag, fp) in enumerate(fps_d.items()):
log.debug('loading %i/%i' % (i + 1, len(fps_d)))
# build/load the children
sWrkr = Scenario(self,
cf_fp=fp,
absolute_fp=self.absolute_fp,
base_dir=os.path.dirname(fp),
tag=tag).setup()
# add to family
assert sWrkr.name not in d, 'scenario \'%s\' already loaded!' % sWrkr.name
d[sWrkr.name] = sWrkr
log.debug('loaded \'%s\'' % sWrkr.name)
self.sWrkr_d = d
log.info('compiled %i scenarios: %s' %
(len(self.sWrkr_d), list(self.sWrkr_d.keys())))
return wdict(self.sWrkr_d)
def cf_compare(
self, #compare control file values between Scenarios
sWrkr_d=None,
logger=None):
if logger is None: logger = self.logger
log = logger.getChild('cf_compare')
if sWrkr_d is None: sWrkr_d = wdict(self.sWrkr_d)
assert len(sWrkr_d) >= 2
#=======================================================================
# collect all the parameters from the children
#=======================================================================
first = True
for childName, sWrkr in sWrkr_d.items():
log.debug('extracting variables from %s' % childName)
#===================================================================
# collect values from this child
#===================================================================
firstC = True
for sectName, svars_d in sWrkr.cfPars_d.items():
if len(svars_d) == 0: continue #skip blank sections
sdf = pd.DataFrame.from_dict(svars_d, orient='index')
sdf.columns = [childName]
#collapse the field names
sdf.index = pd.Series(np.full(len(sdf), sectName)).str.cat(
pd.Series(sdf.index), sep='.')
if firstC:
cdf = sdf
firstC = False
else:
cdf = cdf.append(sdf)
#add the control file path itself
cdf.loc['cf_fp', childName] = sWrkr.cf_fp
#===================================================================
# update library
#===================================================================
if first:
mdf = cdf
first = False
else:
mdf = mdf.join(cdf)
#=======================================================================
# compare values
#=======================================================================
#determine if all values match by row
mdf['compare'] = mdf.eq(other=mdf.iloc[:, 0], axis=0).all(axis=1)
log.info(
'finished w/ %i (of %i) parameters matching between %i scenarios' %
(mdf['compare'].sum(), len(mdf.index), len(mdf.columns)))
return mdf
def riskCurves(
self, #plot a risk curve comparing all the scenarios
sWrkr_d=None, #container of scenario works to plot curve comparison
logger=None,
#plot keys
y1lab='AEP', #yaxis label and plot type c ontrol
#plot formatters
val_str='*no',
**plotKwargs):
#=======================================================================
# defaults
#=======================================================================
if logger is None: logger = self.logger
log = logger.getChild('riskCurves')
if sWrkr_d is None: sWrkr_d = wdict(self.sWrkr_d)
#=======================================================================
# collect data from children
#=======================================================================
plotPars_d = dict()
#loop through each
first = True
for childName, sWrkr in sWrkr_d.items():
log.debug('preping %s' % childName)
plotPars_d[childName] = {
'ttl_df': sWrkr.data_d['r_ttl'],
'ead_tot': sWrkr.ead_tot,
'impStyle_d': sWrkr.impStyle_d.copy(),
}
if first:
self.impact_name = sWrkr.impact_name
first = False
return self.plot_mRiskCurves(plotPars_d,
y1lab=y1lab,
impactFmtFunc=self.impactFmtFunc,
val_str=val_str,
logger=log,
**plotKwargs)
def build_composite(
self, #merge data from a collection of asset models
sWrkr_d=None, #container of scenario works to plot curve comparison
new_wrkr=True, #whether to build a new worker
):
log = self.logger.getChild('collect_ttls')
if sWrkr_d is None: sWrkr_d = wdict(self.sWrkr_d)
#more logicla for single plots
self.name = self.tag
#=======================================================================
# collect data
#=======================================================================
mdf = None
ead_d = dict()
for childName, sWrkr in sWrkr_d.items():
dfi_raw = sWrkr.data_d['r_ttl'].copy() #set by prep_ttl()
dfi = dfi_raw.loc[:, 'impacts'].rename(childName).astype(
float).to_frame()
ead_d[childName] = sWrkr.ead_tot
if mdf is None:
#pull some from the first
self.impact_name = sWrkr.impact_name
self.rtail, self.ltail = sWrkr.rtail, sWrkr.ltail
mindex = pd.MultiIndex.from_frame(
dfi_raw.loc[:, ['aep', 'ari', 'note', 'plot']])
mdf = dfi
else:
mdf = mdf.join(dfi)
#=======================================================================
# build new worker
#=======================================================================
if new_wrkr:
cWrkr = sWrkr.copy(name='%s_%i_composite' %
(self.tag, len(sWrkr_d))) #start with a copy
#===================================================================
# convert the data to standard ttl format
#===================================================================
cttl_df = dfi_raw.drop(['impacts', 'ari'], axis=1).copy()
cttl_df[self.impact_name] = mdf.sum(axis=1)
#fix order
"""
prep_ttl() is column location senstigvite
"""
l = cttl_df.columns.to_list()
l.remove(self.impact_name)
l = [l[0], self.impact_name] + l[1:]
cttl_df = cttl_df.loc[:, l]
#add the ead row at the bottom
cWrkr.data_d['r_ttl'] = cttl_df.append(pd.Series({
'aep':
'ead',
'note':
'integration',
'plot':
False,
self.impact_name:
sum(ead_d.values())
}),
ignore_index=True)
"""this function isnt set up very well.. easier to do it ourselves here
#get just the combined impacts (drop tails)
bx = dfi_raw['note']=='impact_sum'
cttl_df = dfi_raw.loc[bx, 'aep'].to_frame().join(pd.Series(mdf[bx].sum(axis=1), name='impacts'))
cWrkr.data_d['r_ttl'] = dfi_raw #set this for pulling handles
cWrkr.data_d['r_ttl'], ead_tot = cWrkr.get_ttl(cttl_df, logger=log,
cols_include=['note', 'plot'])"""
#=======================================================================
# wrap
#=======================================================================
mdf.index = mindex
"""
view(mdf)
"""
log.info('collected %i' % len(mdf.columns))
self.cdxind = mdf
self.ead_d = ead_d
self.ead_tot = sum(ead_d.values())
return self.cdxind, cWrkr