def test_melt():
pdf = pd.DataFrame({"A": list("abcd") * 5, "B": list("XY") * 10, "C": np.random.randn(20)})
ddf = dd.from_pandas(pdf, 4)
list_eq(dd.melt(ddf), pd.melt(pdf))
list_eq(dd.melt(ddf, id_vars="C"), pd.melt(pdf, id_vars="C"))
list_eq(dd.melt(ddf, value_vars="C"), pd.melt(pdf, value_vars="C"))
list_eq(
dd.melt(ddf, value_vars=["A", "C"], var_name="myvar"), pd.melt(pdf, value_vars=["A", "C"], var_name="myvar")
)
list_eq(
dd.melt(ddf, id_vars="B", value_vars=["A", "C"], value_name="myval"),
pd.melt(pdf, id_vars="B", value_vars=["A", "C"], value_name="myval"),
)
def _stats_plot(self, element, y, data=None):
"""
Helper method to generate element from indexed dataframe.
"""
data, x, y = self._process_args(data, None, y)
opts = {
'plot': dict(self._plot_opts), # labelled=[]),
'norm': self._norm_opts,
'style': self._style_opts
}
if not isinstance(y, (list, tuple)):
ranges = {y: self._dim_ranges['y']}
return (element(
data, self.by,
y).redim.range(**ranges).relabel(**self._relabel).opts(**opts))
labelled = ['y' if self.invert else 'x'
] if self.group_label != 'Group' else []
if self.value_label != 'value':
labelled.append('x' if self.invert else 'y')
opts['plot']['labelled'] = labelled
kdims = [self.group_label]
ranges = {self.value_label: self._dim_ranges['y']}
data = data[list(y)]
if check_library(data, 'dask'):
from dask.dataframe import melt
else:
melt = pd.melt
df = melt(data, var_name=self.group_label, value_name=self.value_label)
return (element(
df, kdims, self.value_label).redim.range(**ranges).redim(
**self._redim).relabel(**self._relabel).opts(**opts))
def _category_plot(self, element, x, y, data):
"""
Helper method to generate element from indexed dataframe.
"""
labelled = ['y' if self.invert else 'x'] if x != 'index' else []
if self.value_label != 'value':
labelled.append('x' if self.invert else 'y')
opts = {
'plot': dict(self._plot_opts, labelled=labelled),
'style': dict(self._style_opts),
'norm': self._norm_opts
}
ranges = {self.value_label: self._dim_ranges['y']}
id_vars = [x]
if any((v in [data.index.names] + ['index']) or v == data.index.name
for v in id_vars):
data = data.reset_index()
data = data[y + [x]]
if check_library(data, 'dask'):
from dask.dataframe import melt
else:
melt = pd.melt
df = melt(data,
id_vars=[x],
var_name=self.group_label,
value_name=self.value_label)
kdims = [x, self.group_label]
return (element(
df, kdims, self.value_label).redim.range(**ranges).redim(
**self._redim).relabel(**self._relabel).opts(**opts))
def test_melt():
pdf = pd.DataFrame({'A': list('abcd') * 5,
'B': list('XY') * 10,
'C': np.random.randn(20)})
ddf = dd.from_pandas(pdf, 4)
list_eq(dd.melt(ddf),
pd.melt(pdf))
list_eq(dd.melt(ddf, id_vars='C'),
pd.melt(pdf, id_vars='C'))
list_eq(dd.melt(ddf, value_vars='C'),
pd.melt(pdf, value_vars='C'))
list_eq(dd.melt(ddf, value_vars=['A', 'C'], var_name='myvar'),
pd.melt(pdf, value_vars=['A', 'C'], var_name='myvar'))
list_eq(dd.melt(ddf, id_vars='B', value_vars=['A', 'C'], value_name='myval'),
pd.melt(pdf, id_vars='B', value_vars=['A', 'C'], value_name='myval'))
def create_diffdiff():
print("\ngenerating spreads vs spreads...")
diff_diff = triu(dateless_df, True)
## merge abbr together, eg. "XXXX (PBF) - XXXX (SSF)" to "XXXX - XXXX (PBF-SSF)"
diff_diff.columns = [
f'{h.split(" (")[0]}{h.split(" (")[1].split(")")[1]} ({h.split(" (")[1].split(")")[0]}-{h.split(" (")[2]}'
for h in diff_diff.columns
]
## verbose version of above's list comprehension
# new_headers = []
# for header in diff_diff.columns:
# split_header = header.split(" (")
# product_A = split_header[0]
# split_section = split_header[1].split(")")
# product_B = split_section[1]
# new_headers.append(f"{product_A}{product_B} ({split_section[0]}-{split_header[2]}")
# diff_diff.columns = new_headers
diff_diff = triu(diff_diff, False)
diff_diff = diff_diff.repartition(npartitions=200)
diff_diff = diff_diff.reset_index(drop=True)
dd_date_col = dd.from_array(date_col)
dd_date_col = dd_date_col.repartition(npartitions=200)
dd_date_col = dd_date_col.reset_index(drop=True)
diff_diff = diff_diff.assign(date=dd_date_col)
diff_diff = dd.melt(
diff_diff,
id_vars="date",
var_name="product_diff",
value_name="price_diff").dropna().reset_index(drop=True)
diff_diff["product_diff"] = diff_diff["product_diff"].astype(
"category")
diff_diff["differential_A"] = diff_diff["product_diff"].str.partition(
" - ")[0]
diff_diff["differential_B"] = diff_diff["product_diff"].str.partition(
" - ")[2]
print(f"\nsaving file... ({round((time.time() - starttime), 2)}s)")
dd.to_csv(df=diff_diff,
filename=os.path.join(os.getcwd(), "cleaned_data",
"diff_diff.csv"),
index=False,
single_file=True,
encoding="utf-8-sig",
chunksize=10000)
print(
f"[diff_diff.csv] saved successfully... ({round((time.time() - starttime), 2)}s)"
)
def calc_enduse(self, eu_fraction_dict, county_energy_dd, temps=False):
"""
Returns Dask DataFrame
"""
unitname_eu_dict = {
'Process Heating': [
'furnace', 'kiln', 'dryer', 'heater', 'oven', 'calciner',
'stove', 'htr', 'furn', 'cupola'
],
'Conventional Boiler Use': ['boiler'],
'CHP and/or Cogeneration Process': ['turbine'],
'Facility HVAC': ['building heat', 'space heater'],
'Machine Drive': ['engine', 'compressor', 'pump', 'rice'],
'Conventional Electricity Generation': ['generator'],
'Other Nonprocess Use': [
'hot water', 'crane', 'water heater', 'comfort heater', 'RTO',
'TODF', 'oxidizer', 'RCO'
]
}
unittype_eu_dict = {
'Process Heating': [
'F', 'PD', 'K', 'PRH', 'O', 'NGLH', 'CF', 'HMH', 'C', 'HPPU',
'CatH', 'COB', 'FeFL', 'Chemical Recovery Furnace', 'IFCE',
'Pulp Mill Lime Kiln', 'Lime Kiln',
'Chemical Recovery Combustion Unit',
'Direct Reduction Furnace', 'Sulfur Recovery Plant'
],
'Conventional Boiler Use': [
'OB', 'S', 'PCWW', 'BFB', 'PCWD', 'PCT', 'CFB', 'PCO', 'OFB',
'PFB'
],
'CHP and/or Cogeneration Process': ['CCCT', 'SCCT'],
'Facility HVAC': ['CH'],
'Other Nonprocess Use':
['HWH', 'TODF', 'ICI', 'FLR', 'RTO', 'II', 'MWC', 'Flare', 'RCO'],
'Conventional Electricity Generation':
['RICE', 'Electricity Generator']
}
def eu_dict_to_df(eu_dict):
"""
Convert unit type/unit name dictionaries to dataframes.
"""
eu_df = pd.DataFrame.from_dict(eu_dict,
orient='index').reset_index()
eu_df = pd.melt(eu_df, id_vars='index',
value_name='unit').rename(columns={
'index': 'end_use'
}).drop('variable', axis=1)
eu_df = eu_df.dropna().set_index('unit')
return eu_df
def eu_unit_type(unit_type, unittype_eu_df):
"""
Match GHGRP unit type to end use specified in unittype_eu_dict.
"""
enduse = re.match('(\w+) \(', unit_type)
if enduse != None:
enduse = re.match('(\w+)', enduse.group())[0]
if enduse in unittype_eu_df.index:
enduse = unittype_eu_df.loc[enduse, 'end_use']
else:
enduse = np.nan
else:
if unit_type in unittype_eu_df.index:
enduse = unittype_eu_df.loc[unit_type, 'end_use']
return enduse
def eu_unit_name(unit_name, unitname_eu_df):
"""
Find keywords in GHGRP unit name descriptions and match them
to appropriate end uses based on unitname_eu_dict.
"""
for i in unitname_eu_df.index:
enduse = re.search(i, unit_name.lower())
if enduse == None:
continue
else:
enduse = unitname_eu_df.loc[i, 'end_use']
return enduse
enduse = np.nan
return enduse
unittype_eu_df = eu_dict_to_df(unittype_eu_dict)
unitname_eu_df = eu_dict_to_df(unitname_eu_dict)
# Base ghgrp energy end use disaggregation on reported unit type and
# unit name.
eu_ghgrp = self.energy_eia923.copy(deep=True)
# First match end uses to provided unit types. Most unit types are
# specified as OCS (other combustion source).
unit_types = eu_ghgrp.UNIT_TYPE.dropna().unique()
type_match = list()
for utype in unit_types:
enduse = eu_unit_type(utype, unittype_eu_df)
type_match.append([utype, enduse])
type_match = pd.DataFrame(type_match, columns=['UNIT_TYPE', 'end_use'])
eu_ghgrp = pd.merge(eu_ghgrp, type_match, on='UNIT_TYPE', how='left')
# Next, match end use by unit name for facilites that report OCS for
# unit type.
eu_ocs = eu_ghgrp[
(eu_ghgrp.UNIT_TYPE == 'OCS (Other combustion source)') |
(eu_ghgrp.UNIT_TYPE.isnull())][['UNIT_TYPE', 'UNIT_NAME']]
eu_ocs['end_use'] = eu_ocs.UNIT_NAME.apply(
lambda x: eu_unit_name(x, unitname_eu_df))
eu_ghgrp.end_use.update(eu_ocs.end_use)
eu_ghgrp.drop(eu_ghgrp.columns.difference(
set([
'COUNTY_FIPS', 'MECS_Region', 'MMBtu_TOTAL', 'MECS_FT',
'PRIMARY_NAICS_CODE', 'MECS_NAICS', 'end_use', 'FACILITY_ID'
])),
axis=1,
inplace=True)
# sum energy of unit types and unit names matched to an end use
eu_ghgrp_matched = eu_ghgrp[eu_ghgrp.end_use.notnull()].pivot_table(
values='MMBtu_TOTAL',
columns='end_use',
index=[
'MECS_Region', 'COUNTY_FIPS', 'PRIMARY_NAICS_CODE',
'MECS_NAICS', 'MECS_FT'
],
aggfunc='sum',
fill_value=0)
eu_ghgrp_matched = eu_ghgrp_matched.join(
eu_ghgrp.pivot_table(values='FACILITY_ID',
index=[
'MECS_Region', 'COUNTY_FIPS',
'PRIMARY_NAICS_CODE', 'MECS_NAICS',
'MECS_FT'
],
aggfunc='count'))
# Calculate the remaining GHGRP facilities energy use
# with MECS data.
eu_ghgrp_notmatched = \
eu_ghgrp[(eu_ghgrp.end_use.isnull()) &
(eu_ghgrp.MECS_FT.notnull())].copy(deep=True)
enduses = eu_fraction_dict['GHGRP'].columns.values
eu_ghgrp_notmatched = pd.merge(eu_ghgrp_notmatched.set_index(
['MECS_NAICS', 'MECS_FT']),
eu_fraction_dict['GHGRP'],
left_index=True,
right_index=True,
how='left')
for eu in enduses:
eu_ghgrp_notmatched[eu] = \
eu_ghgrp_notmatched.MMBtu_TOTAL.multiply(
eu_ghgrp_notmatched[eu], fill_value=0
)
agg_cols = [x for x in itools.product(enduses, ['sum'])]
agg_cols.append(('FACILITY_ID', 'count'))
eu_ghgrp_notmatched = eu_ghgrp_notmatched.reset_index().groupby(
[
'MECS_Region',
'COUNTY_FIPS',
'PRIMARY_NAICS_CODE',
'MECS_NAICS',
'MECS_FT',
],
as_index=False).agg(dict(agg_cols))
eu_ghgrp_notmatched.set_index('MECS_NAICS', inplace=True)
eu_ghgrp_matched.reset_index(
['MECS_Region', 'COUNTY_FIPS', 'PRIMARY_NAICS_CODE', 'MECS_FT'],
inplace=True)
for df in [eu_ghgrp_matched, eu_ghgrp_notmatched]:
df.rename(columns={
'PRIMARY_NAICS_CODE': 'NAICS',
'FACILITY_ID': 'est_count'
},
inplace=True)
df['data_source'] = 'ghgrp'
# Calculate end use of energy estimated from MECS data with MECS end
# use.
enduses = eu_fraction_dict['nonGHGRP'].columns.values
eu_energy_dd = dd.merge(
county_energy_dd[county_energy_dd.data_source == 'mecs_ipf'],
eu_fraction_dict['nonGHGRP'].reset_index('MECS_FT'),
on=['MECS_NAICS', 'MECS_FT'],
how='left')
for eu in enduses:
eu_energy_dd[eu] = \
eu_energy_dd.MMBtu_TOTAL.mul(eu_energy_dd[eu],
fill_value=0)
# This throws FutureWanring related to sorting for pandas concat,
# but currently there's no option to address this in dd.concat
eu_energy_dd = dd.concat([
df for df in [eu_energy_dd, eu_ghgrp_matched, eu_ghgrp_notmatched]
],
axis='index',
join='outer',
interleave_partitions=True)
eu_energy_dd_final = dd.melt(eu_energy_dd.reset_index(),
value_vars=enduses.tolist(),
id_vars=[
'MECS_NAICS', 'COUNTY_FIPS',
'Emp_Size', 'MECS_FT', 'MECS_Region',
'data_source', 'est_count', 'fipscty',
'fipstate', 'NAICS'
],
var_name='End_use',
value_name='MMBtu')
# clean up by removing MMBtu values == 0..
eu_energy_dd_final = \
eu_energy_dd_final[eu_energy_dd_final.MMBtu !=0]
eu_energy_dd_final = eu_energy_dd_final.set_index('MECS_NAICS')
# compute=True)
return eu_energy_dd_final
ssrt_metrics['NARGUID'] = ssrt_metrics['filename'].apply(
lambda x: x.split('_')[-1].replace('.csv', ''), meta=str)
ssrt_metrics['Underlying Distribution'] = ssrt_metrics['filename'].apply(
lambda x: '_'.join(x.split('/')[-1].split('_')[:-1]), meta=str)
ssrt_metrics = ssrt_metrics.drop('filename', axis=1)
ssrt_metrics['graded_both'] = ssrt_metrics['SSRT_w_graded']
ssrt_metrics = ssrt_metrics.rename(
columns={
'SSRT': 'standard',
'SSRT_w_guesses': 'guesses',
'SSRT_w_graded': 'graded_go'
})
print('melting...')
melt_df = dd.melt(
ssrt_metrics,
id_vars=['SSD', 'Underlying Distribution'],
value_vars=['standard', 'guesses', 'graded_go', 'graded_both'],
var_name='Assumed Distribution',
value_name='SSRT')
# Formal Names
renaming_map = {
'standard': 'Independent Race',
'guesses': 'Guessing',
'graded_go': 'Slowed Go Processing',
'graded_both': 'Confusion',
'ABCD data': 'ABCD Data'
}
melt_df["Assumed Distribution"] = melt_df["Assumed Distribution"].replace(
renaming_map)
melt_df["Underlying Distribution"] = melt_df[
"Underlying Distribution"].replace(renaming_map)
