def parsepars(parstring):
""" parse parameters specific to the ACEnsemble parstring """
import numpy as np
pars = {}
sectiontitles = ["START OF DATA:", "START OF AC:", "START OF ENS:"]
startidxs, finidxs = parstr_sectionindexes(parstring, sectiontitles)
alllines = parstring.splitlines()
# parse the data section --------------------------------------------------------
lines = alllines[startidxs[0]:finidxs[0]]
# data
pars['datafile'] = lines[0].split()[0]
try:
data = read_gslib(pars['datafile'])
except:
print("ERROR: `{}` must be geoeas-formatted!".format(pars['datafile']))
raise
cols = data.columns
# locations
dh, ifx, ify, ifz = [int(i) for i in lines[1].split()[:4]]
assert all(c > 0 for c in [ifx, ify]), "ERROR: ifx and ify cols must be > zero"
pars['locations'] = data[[cols[ifx - 1], cols[ify - 1]]].values
if ifz > 0:
pars['locations'] = np.c_[pars['locations'], data[cols[ifz - 1]].values]
pars['dhs'] = dh[cols[dh - 1]] if dh > 0 else None
# mvdata
varcols = [int(c) for c in lines[2][:lines[2].rfind('-')].split()]
pars['mvdata'] = data[[cols[vc - 1] for vc in varcols]].values
# output options
pars['outfile'] = lines[3].split()[0]
pars['saveall'], pars['recode'] = [bool(int(p)) for p in lines[4].split()[:2]]
pars['appendout'] = bool(int(lines[5].split()[0]))
# parse the ac section --------------------------------------------------------
lines = alllines[startidxs[1]:finidxs[1]]
# search stuff
pars['nnears'] = int(lines[0].split()[0])
angs = [float(a) for a in lines[1].split()[:3]]
ranges = [float(r) for r in lines[2].split()[:3]]
pars['searchparams'] = tuple(angs + ranges)
# cluster stuff, metrics and proportions
pars['cluster_method'] = lines[3].split()[0]
pars['acmetric'] = [int(a) for a in lines[4][:lines[4].rfind('-')].split()]
pars['acprop'] = [float(a) for a in lines[5][:lines[4].rfind('-')].split()]
# parse the ens section --------------------------------------------------------
lines = alllines[startidxs[2]:finidxs[2]]
# rseed
pars['rseed'] = int(lines[0].split()[0])
pars['nreal'] = int(lines[1].split()[0])
pars['minvars'] = int(lines[2].split()[0])
pars['fnclus'], pars['tnclus'] = [int(c) for c in lines[3].split()[:2]]
pars['minremove'], pars['maxremove'] = [float(c) for c in lines[4].split()[:2]]
pars['minfound'], pars['maxfound'] = [float(c) for c in lines[5].split()[:2]]
pars['consensus_method'] = lines[6].split()[0]
return pars
def main():
# some defaults to make parts of this generic
cluster_object = sp.ACEnsemble
thisname = cluster_object.__name__
shortname = thisname[:5].lower()
# setup the argparser and check for no args
parser = ArgumentParser(description="Runner for {}".format(thisname))
parser.add_argument('parfile', type=str, nargs='?',
help="the parfile to call `{}` with".format(thisname))
args = parser.parse_args()
if args.parfile is None: # if nothing passed, print the default parfile
writeparfile(defaultparstr, '{}.par'.format(thisname.lower()))
sys.exit(0)
else:
assert os.path.isfile(args.parfile), "ERROR: {} does not exist!".format(args.parfile)
# parse the parameter file
pars = parsepars(readparfile(args.parfile))
# collect pars that are not sent to the cluster object init
datafile = pars.pop('datafile')
outfile = pars.pop('outfile')
saveall = pars.pop('saveall')
recode = pars.pop('recode')
appendout = pars.pop('appendout')
fnclus = pars.pop('fnclus')
tnclus = pars.pop('tnclus')
# nthread = pars.pop('nthread')
consensus_method = pars.pop('consensus_method')
# setup the output datafile
if appendout:
data = read_gslib(datafile)
else:
from pandas import DataFrame
data = DataFrame()
# generate the clustering object
model = cluster_object(**pars)
model.fit(tnclus)
labels = model.predict(fnclus, method=consensus_method)
data['{}_clusters'.format(shortname)] = labels
if saveall:
if recode:
clusterings, _ = sp.reclass_clusters(labels, model.clusterings)
else:
clusterings = model.clusterings
for i in range(pars['nreal']):
data['{}_real{}'.format(shortname, i)] = clusterings[:, i]
# write the output
write_gslib(data, outfile)
def parsepars(parstring):
""" parse parameters specific to the ACEnsemble parstring """
import numpy as np
pars = {}
sectiontitles = ["START OF DATA:", "START OF AC:"]
startidxs, finidxs = parstr_sectionindexes(parstring, sectiontitles)
alllines = parstring.splitlines()
# parse the data section --------------------------------------------------------
lines = alllines[startidxs[0]:finidxs[0]]
# data
pars['datafile'] = lines[0].split()[0]
try:
data = read_gslib(pars['datafile'])
except:
print("ERROR: `{}` must be geoeas-formatted!".format(pars['datafile']))
raise
cols = data.columns
# locations
ifx, ify, ifz = [int(i) for i in lines[1].split()[:3]]
assert all(c > 0
for c in [ifx, ify]), "ERROR: ifx and ify cols must be > zero"
pars['locations'] = data[[cols[ifx - 1], cols[ify - 1]]].values
if ifz > 0:
pars['locations'] = np.c_[pars['locations'],
data[cols[ifz - 1]].values]
# mvdata
varcols = [int(c) for c in lines[2][:lines[2].rfind('-')].split()]
pars['mvdata'] = data[[cols[vc - 1] for vc in varcols]].values
# output options
pars['outfile'] = lines[3].split()[0]
pars['appendout'] = bool(int(lines[4].split()[0]))
# parse the ac section --------------------------------------------------------
lines = alllines[startidxs[1]:finidxs[1]]
# search stuff
pars['nnears'] = int(lines[0].split()[0])
angs = [float(a) for a in lines[1].split()[:3]]
ranges = [float(r) for r in lines[2].split()[:3]]
pars['searchparams'] = tuple(angs + ranges)
# cluster stuff, metrics and proportions
pars['cluster_method'] = lines[3].split()[0]
pars['acmetric'] = sp.accluster._acmetric_MAPPING_R[int(
lines[4].split()[0])]
pars['nclus'] = int(lines[5].split()[0])
return pars
def main():
# some defaults to make parts of this generic
cluster_object = sp.ACCluster
thisname = cluster_object.__name__
shortname = thisname[:3].lower()
# setup the argparser and check for no args
parser = ArgumentParser(description="Runner for {}".format(thisname))
parser.add_argument('parfile',
type=str,
nargs='?',
help="the parfile to call `{}` with".format(thisname))
args = parser.parse_args()
if args.parfile is None: # if nothing passed, print the default parfile
writeparfile(defaultparstr, '{}.par'.format(thisname.lower()))
sys.exit(0)
else:
assert os.path.isfile(
args.parfile), "ERROR: {} does not exist!".format(args.parfile)
# parse the parameter file
pars = parsepars(readparfile(args.parfile))
# collect pars that are not sent to the cluster object init
datafile = pars.pop('datafile')
outfile = pars.pop('outfile')
appendout = pars.pop('appendout')
# setup the output datafile
if appendout:
data = read_gslib(datafile)
else:
from pandas import DataFrame
data = DataFrame()
# generate the clustering object
model = cluster_object(**pars)
model.fit()
labels = model.predict()
data['{}_clusters'.format(shortname)] = labels
# write the output
write_gslib(data, outfile)