def get_tvpb_logsum(self, path_type, orig, dest, tod, demographic_segment, want_choices, trace_label=None):
# assume they have given us a more specific name (since there may be more than one active wrapper)
trace_label = trace_label or 'get_tvpb_logsum'
trace_label = tracing.extend_trace_label(trace_label, path_type)
recipe = 'tour_mode_choice'
with chunk.chunk_log(trace_label):
logsum_df = \
self.build_virtual_path(recipe, path_type, orig, dest, tod, demographic_segment,
want_choices=want_choices, trace_label=trace_label)
trace_hh_id = inject.get_injectable("trace_hh_id", None)
if trace_hh_id:
filter_targets = tracing.trace_targets(orig)
# choices from preceding run (because random numbers)
override_choices = logsum_df['path_num'] if want_choices else None
if filter_targets.any():
self.build_virtual_path(recipe, path_type, orig, dest, tod, demographic_segment,
want_choices=want_choices, override_choices=override_choices,
trace_label=trace_label, filter_targets=filter_targets, trace=True)
return logsum_df
def compute_columns(df, model_settings, configs_dir, trace_label=None):
"""
Evaluate expressions_spec in context of df, with optional additional pipeline tables in locals
Parameters
----------
df : pandas DataFrame
or if None, expect name of pipeline table to be specified by DF in model_settings
model_settings : dict or str
dict with keys:
DF - df_alias and (additionally, if df is None) name of pipeline table to load as df
SPEC - name of expressions file (csv suffix optional) if different from model_settings
TABLES - list of pipeline tables to load and make available as (read only) locals
str:
name of yaml file in confirs_dir to load dict from
configs_dir
trace_label
Returns
-------
results: pandas.DataFrame
one column for each expression (except temps with ALL_CAP target names)
same index as df
"""
if isinstance(model_settings, str):
model_settings_name = model_settings
model_settings = config.read_model_settings(configs_dir, '%s.yaml' % model_settings)
assert model_settings, "Found no model settings for %s" % model_settings_name
else:
model_settings_name = 'dict'
assert 'DF' in model_settings, \
"Expected to find 'DF' in %s" % model_settings_name
df_name = model_settings.get('DF')
helper_table_names = model_settings.get('TABLES', [])
expressions_spec_name = model_settings.get('SPEC', model_settings_name)
assert expressions_spec_name is not None, \
"Expected to find 'SPEC' in %s" % model_settings_name
if trace_label is None:
trace_label = expressions_spec_name
if not expressions_spec_name.endswith(".csv"):
expressions_spec_name = '%s.csv' % expressions_spec_name
expressions_spec = assign.read_assignment_spec(os.path.join(configs_dir, expressions_spec_name))
tables = {t: inject.get_table(t).to_frame() for t in helper_table_names}
# if df was passed in, df might be a slice, or any other table, but DF is it's local alias
assert df_name not in tables, "Did not expect to find df '%s' in TABLES" % df_name
tables[df_name] = df
locals_dict = local_utilities()
locals_dict.update(tables)
results, trace_results, trace_assigned_locals \
= assign.assign_variables(expressions_spec,
df,
locals_dict,
trace_rows=tracing.trace_targets(df))
if trace_results is not None:
tracing.trace_df(trace_results,
label=trace_label,
slicer='NONE',
warn_if_empty=True)
if trace_assigned_locals:
tracing.write_csv(trace_assigned_locals, file_name="%s_locals" % trace_label)
return results
def choose_MAZ_for_TAZ(taz_sample, MAZ_size_terms, trips, network_los,
alt_dest_col_name, trace_label):
"""
Convert taz_sample table with TAZ zone sample choices to a table with a MAZ zone chosen for each TAZ
choose MAZ probabilistically (proportionally by size_term) from set of MAZ zones in parent TAZ
Parameters
----------
taz_sample: dataframe with duplicated index <chooser_id_col> and columns: <alt_dest_col_name>, prob, pick_count
MAZ_size_terms: dataframe with duplicated index <chooser_id_col> and columns: zone_id, dest_TAZ, size_term
Returns
-------
dataframe with with duplicated index <chooser_id_col> and columns: <alt_dest_col_name>, prob, pick_count
"""
if len(taz_sample) == 0:
# it can happen that all trips have no viable destinations (and so are dropped from the sample)
# in which case we can just return the empty taz_sample, since it has the same columns
return taz_sample.copy()
# we had to use alt_dest_col_name as specified in model_settings for interaction_sample
# because expressions reference it to look up size_terms by trip purpose
DEST_MAZ = alt_dest_col_name
DEST_TAZ = f"{alt_dest_col_name}_TAZ"
taz_sample.rename(columns={alt_dest_col_name: DEST_TAZ}, inplace=True)
trace_hh_id = inject.get_injectable("trace_hh_id", None)
have_trace_targets = trace_hh_id and tracing.has_trace_targets(taz_sample)
if have_trace_targets:
trace_label = tracing.extend_trace_label(trace_label,
'choose_MAZ_for_TAZ')
# write taz choices, pick_counts, probs
trace_targets = tracing.trace_targets(taz_sample)
tracing.trace_df(taz_sample[trace_targets],
label=tracing.extend_trace_label(
trace_label, 'taz_sample'),
transpose=False)
# print(f"taz_sample\n{taz_sample}")
# alt_dest_TAZ prob pick_count
# trip_id
# 4343721 12 0.000054 1
# 4343721 20 0.001864 2
taz_choices = taz_sample[[DEST_TAZ, 'prob']].reset_index(drop=False)
taz_choices = taz_choices.reindex(
taz_choices.index.repeat(taz_sample.pick_count)).reset_index(drop=True)
taz_choices = taz_choices.rename(columns={'prob': 'TAZ_prob'})
# print(f"taz_choices\n{taz_choices}")
# trip_id alt_dest_TAZ prob
# 0 4343721 12 0.000054
# 1 4343721 20 0.001864
# 2 4343721 20 0.001864
# print(f"MAZ_size_terms\n{MAZ_size_terms.df}")
# work escort shopping eatout othmaint social othdiscr univ
# alt_dest
# 2 31.0 9.930 0.042 0.258 0.560 0.520 10.856 0.042
# 3 0.0 3.277 0.029 0.000 0.029 0.029 7.308 0.029
# 4 0.0 1.879 0.023 0.000 0.023 0.023 5.796 0.023
# just to make it clear we are siloing choices by chooser_id
chooser_id_col = taz_sample.index.name # should be canonical chooser index name (e.g. 'trip_id')
# for random_for_df, we need df with de-duplicated chooser canonical index
chooser_df = pd.DataFrame(
index=taz_sample.index[~taz_sample.index.duplicated()])
num_choosers = len(chooser_df)
assert chooser_df.index.name == chooser_id_col
# to make choices, <taz_sample_size> rands for each chooser (one rand for each sampled TAZ)
# taz_sample_size will be model_settings['SAMPLE_SIZE'] samples, except if we are estimating
taz_sample_size = taz_choices.groupby(
chooser_id_col)[DEST_TAZ].count().max()
# taz_choices index values should be contiguous
assert (taz_choices[chooser_id_col] == np.repeat(chooser_df.index,
taz_sample_size)).all()
# we need to choose a MAZ for each DEST_TAZ choice
# probability of choosing MAZ based on MAZ size_term fraction of TAZ total
# there will be a different set (and number) of candidate MAZs for each TAZ
# (preserve index, which will have duplicates as result of join)
maz_taz = network_los.maz_taz_df[['MAZ', 'TAZ']].rename(columns={
'TAZ': DEST_TAZ,
'MAZ': DEST_MAZ
})
maz_sizes = pd.merge(taz_choices[[chooser_id_col, DEST_TAZ]].reset_index(),
maz_taz,
how='left',
on=DEST_TAZ).set_index('index')
purpose = maz_sizes['trip_id'].map(
trips.purpose) # size term varies by purpose
maz_sizes['size_term'] = MAZ_size_terms.get(maz_sizes[DEST_MAZ], purpose)
# print(f"maz_sizes\n{maz_sizes}")
# trip_id alt_dest_TAZ alt_dest size_term
# index
# 0 4343721 12 3445 0.019
# 0 4343721 12 11583 0.017
# 0 4343721 12 21142 0.020
if have_trace_targets:
# write maz_sizes: maz_sizes[index,trip_id,dest_TAZ,zone_id,size_term]
maz_sizes_trace_targets = tracing.trace_targets(maz_sizes,
slicer='trip_id')
trace_maz_sizes = maz_sizes[maz_sizes_trace_targets]
tracing.trace_df(trace_maz_sizes,
label=tracing.extend_trace_label(
trace_label, 'maz_sizes'),
transpose=False)
# number of DEST_TAZ candidates per chooser
maz_counts = maz_sizes.groupby(maz_sizes.index).size().values
# print(maz_counts)
# max number of MAZs for any TAZ
max_maz_count = maz_counts.max()
# print(f"max_maz_count {max_maz_count}")
# offsets of the first and last rows of each chooser in sparse interaction_utilities
last_row_offsets = maz_counts.cumsum()
first_row_offsets = np.insert(last_row_offsets[:-1], 0, 0)
# repeat the row offsets once for each dummy utility to insert
# (we want to insert dummy utilities at the END of the list of alternative utilities)
# inserts is a list of the indices at which we want to do the insertions
inserts = np.repeat(last_row_offsets, max_maz_count - maz_counts)
# insert zero filler to pad each alternative set to same size
padded_maz_sizes = np.insert(maz_sizes.size_term.values, inserts, 0.0)
padded_maz_sizes = padded_maz_sizes.reshape(-1, max_maz_count)
# prob array with one row TAZ_choice, one column per alternative
row_sums = padded_maz_sizes.sum(axis=1)
maz_probs = np.divide(padded_maz_sizes, row_sums.reshape(-1, 1))
assert maz_probs.shape == (num_choosers * taz_sample_size, max_maz_count)
rands = pipeline.get_rn_generator().random_for_df(
chooser_df, n=taz_sample_size).reshape(-1, 1)
assert len(rands) == num_choosers * taz_sample_size
assert len(rands) == maz_probs.shape[0]
# make choices
# positions is array with the chosen alternative represented as a column index in probs
# which is an integer between zero and max_maz_count
positions = np.argmax((maz_probs.cumsum(axis=1) - rands) > 0.0, axis=1)
# shouldn't have chosen any of the dummy pad positions
assert (positions < maz_counts).all()
taz_choices[DEST_MAZ] = maz_sizes[DEST_MAZ].take(positions +
first_row_offsets)
taz_choices['MAZ_prob'] = maz_probs[np.arange(maz_probs.shape[0]),
positions]
taz_choices['prob'] = taz_choices['TAZ_prob'] * taz_choices['MAZ_prob']
if have_trace_targets:
taz_choices_trace_targets = tracing.trace_targets(taz_choices,
slicer='trip_id')
trace_taz_choices_df = taz_choices[taz_choices_trace_targets]
tracing.trace_df(trace_taz_choices_df,
label=tracing.extend_trace_label(
trace_label, 'taz_choices'),
transpose=False)
lhs_df = trace_taz_choices_df[['trip_id', DEST_TAZ]]
alt_dest_columns = [f'dest_maz_{c}' for c in range(max_maz_count)]
# following the same logic as the full code, but for trace cutout
trace_maz_counts = maz_counts[taz_choices_trace_targets]
trace_last_row_offsets = maz_counts[taz_choices_trace_targets].cumsum()
trace_inserts = np.repeat(trace_last_row_offsets,
max_maz_count - trace_maz_counts)
# trace dest_maz_alts
padded_maz_sizes = np.insert(trace_maz_sizes[DEST_MAZ].values,
trace_inserts,
0.0).reshape(-1, max_maz_count)
df = pd.DataFrame(data=padded_maz_sizes,
columns=alt_dest_columns,
index=trace_taz_choices_df.index)
df = pd.concat([lhs_df, df], axis=1)
tracing.trace_df(df,
label=tracing.extend_trace_label(
trace_label, 'dest_maz_alts'),
transpose=False)
# trace dest_maz_size_terms
padded_maz_sizes = np.insert(trace_maz_sizes['size_term'].values,
trace_inserts,
0.0).reshape(-1, max_maz_count)
df = pd.DataFrame(data=padded_maz_sizes,
columns=alt_dest_columns,
index=trace_taz_choices_df.index)
df = pd.concat([lhs_df, df], axis=1)
tracing.trace_df(df,
label=tracing.extend_trace_label(
trace_label, 'dest_maz_size_terms'),
transpose=False)
# trace dest_maz_probs
df = pd.DataFrame(data=maz_probs[taz_choices_trace_targets],
columns=alt_dest_columns,
index=trace_taz_choices_df.index)
df = pd.concat([lhs_df, df], axis=1)
df['rand'] = rands[taz_choices_trace_targets]
tracing.trace_df(df,
label=tracing.extend_trace_label(
trace_label, 'dest_maz_probs'),
transpose=False)
taz_choices = taz_choices.drop(columns=['TAZ_prob', 'MAZ_prob'])
taz_choices = taz_choices.groupby([chooser_id_col, DEST_MAZ
]).agg(prob=('prob', 'max'),
pick_count=('prob', 'count'))
taz_choices.reset_index(level=DEST_MAZ, inplace=True)
return taz_choices
def choose_MAZ_for_TAZ(taz_sample, MAZ_size_terms, trace_label):
"""
Convert taz_sample table with TAZ zone sample choices to a table with a MAZ zone chosen for each TAZ
choose MAZ probabilistically (proportionally by size_term) from set of MAZ zones in parent TAZ
Parameters
----------
taz_sample: dataframe with duplicated index <chooser_id_col> and columns: <DEST_TAZ>, prob, pick_count
MAZ_size_terms: dataframe with duplicated index <chooser_id_col> and columns: zone_id, dest_TAZ, size_term
Returns
-------
dataframe with with duplicated index <chooser_id_col> and columns: <DEST_MAZ>, prob, pick_count
"""
# print(f"taz_sample\n{taz_sample}")
# dest_TAZ prob pick_count person_id
# tour_id
# 542963 18 0.004778 1 13243
# 542963 53 0.004224 2 13243
# 542963 59 0.008628 1 13243
trace_hh_id = inject.get_injectable("trace_hh_id", None)
have_trace_targets = trace_hh_id and tracing.has_trace_targets(taz_sample)
if have_trace_targets:
trace_label = tracing.extend_trace_label(trace_label,
'choose_MAZ_for_TAZ')
CHOOSER_ID = taz_sample.index.name # zone_id for tours, but person_id for location choice
assert CHOOSER_ID is not None
# write taz choices, pick_counts, probs
trace_targets = tracing.trace_targets(taz_sample)
tracing.trace_df(taz_sample[trace_targets],
label=tracing.extend_trace_label(
trace_label, 'taz_sample'),
transpose=False)
# redupe taz_sample[[DEST_TAZ, 'prob']] using pick_count to repeat rows
taz_choices = taz_sample[[DEST_TAZ, 'prob']].reset_index(drop=False)
taz_choices = taz_choices.reindex(
taz_choices.index.repeat(taz_sample.pick_count)).reset_index(drop=True)
taz_choices = taz_choices.rename(columns={'prob': 'TAZ_prob'})
# print(f"taz_choices\n{taz_choices}")
# tour_id dest_TAZ TAZ_prob
# 0 542963 18 0.004778
# 1 542963 53 0.004224
# 2 542963 53 0.004224
# 3 542963 59 0.008628
# print(f"MAZ_size_terms\n{MAZ_size_terms}")
# zone_id dest_TAZ size_term
# 0 6097 2 7.420
# 1 16421 2 9.646
# 2 24251 2 10.904
# just to make it clear we are siloing choices by chooser_id
chooser_id_col = taz_sample.index.name # should be canonical chooser index name (e.g. 'person_id')
# for random_for_df, we need df with de-duplicated chooser canonical index
chooser_df = pd.DataFrame(
index=taz_sample.index[~taz_sample.index.duplicated()])
num_choosers = len(chooser_df)
assert chooser_df.index.name == chooser_id_col
# to make choices, <taz_sample_size> rands for each chooser (one rand for each sampled TAZ)
# taz_sample_size will be model_settings['SAMPLE_SIZE'] samples, except if we are estimating
taz_sample_size = taz_choices.groupby(
chooser_id_col)[DEST_TAZ].count().max()
# taz_choices index values should be contiguous
assert (taz_choices[chooser_id_col] == np.repeat(chooser_df.index,
taz_sample_size)).all()
# we need to choose a MAZ for each DEST_TAZ choice
# probability of choosing MAZ based on MAZ size_term fraction of TAZ total
# there will be a different set (and number) of candidate MAZs for each TAZ
# (preserve index, which will have duplicates as result of join)
# maz_sizes.index is the integer offset into taz_choices of the taz for which the maz_size row is a candidate)
maz_sizes = pd.merge(taz_choices[[chooser_id_col, DEST_TAZ]].reset_index(),
MAZ_size_terms,
how='left',
on=DEST_TAZ).set_index('index')
# tour_id dest_TAZ zone_id size_term
# index
# 0 542963 18 498 12.130
# 0 542963 18 7696 18.550
# 0 542963 18 15431 8.678
# 0 542963 18 21429 29.938
# 1 542963 53 17563 34.252
if have_trace_targets:
# write maz_sizes: maz_sizes[index,tour_id,dest_TAZ,zone_id,size_term]
maz_sizes_trace_targets = tracing.trace_targets(maz_sizes,
slicer=CHOOSER_ID)
trace_maz_sizes = maz_sizes[maz_sizes_trace_targets]
tracing.trace_df(trace_maz_sizes,
label=tracing.extend_trace_label(
trace_label, 'maz_sizes'),
transpose=False)
# number of DEST_TAZ candidates per chooser
maz_counts = maz_sizes.groupby(maz_sizes.index).size().values
# max number of MAZs for any TAZ
max_maz_count = maz_counts.max()
# offsets of the first and last rows of each chooser in sparse interaction_utilities
last_row_offsets = maz_counts.cumsum()
first_row_offsets = np.insert(last_row_offsets[:-1], 0, 0)
# repeat the row offsets once for each dummy utility to insert
# (we want to insert dummy utilities at the END of the list of alternative utilities)
# inserts is a list of the indices at which we want to do the insertions
inserts = np.repeat(last_row_offsets, max_maz_count - maz_counts)
# insert zero filler to pad each alternative set to same size
padded_maz_sizes = np.insert(maz_sizes.size_term.values, inserts,
0.0).reshape(-1, max_maz_count)
# prob array with one row TAZ_choice, one column per alternative
row_sums = padded_maz_sizes.sum(axis=1)
maz_probs = np.divide(padded_maz_sizes, row_sums.reshape(-1, 1))
assert maz_probs.shape == (num_choosers * taz_sample_size, max_maz_count)
rands = pipeline.get_rn_generator().random_for_df(chooser_df,
n=taz_sample_size)
rands = rands.reshape(-1, 1)
assert len(rands) == num_choosers * taz_sample_size
assert len(rands) == maz_probs.shape[0]
# make choices
# positions is array with the chosen alternative represented as a column index in probs
# which is an integer between zero and max_maz_count
positions = np.argmax((maz_probs.cumsum(axis=1) - rands) > 0.0, axis=1)
# shouldn't have chosen any of the dummy pad positions
assert (positions < maz_counts).all()
taz_choices[DEST_MAZ] = maz_sizes['zone_id'].take(positions +
first_row_offsets)
taz_choices['MAZ_prob'] = maz_probs[np.arange(maz_probs.shape[0]),
positions]
taz_choices['prob'] = taz_choices['TAZ_prob'] * taz_choices['MAZ_prob']
if have_trace_targets:
taz_choices_trace_targets = tracing.trace_targets(taz_choices,
slicer=CHOOSER_ID)
trace_taz_choices_df = taz_choices[taz_choices_trace_targets]
tracing.trace_df(trace_taz_choices_df,
label=tracing.extend_trace_label(
trace_label, 'taz_choices'),
transpose=False)
lhs_df = trace_taz_choices_df[[CHOOSER_ID, DEST_TAZ]]
alt_dest_columns = [f'dest_maz_{c}' for c in range(max_maz_count)]
# following the same logic as the full code, but for trace cutout
trace_maz_counts = maz_counts[taz_choices_trace_targets]
trace_last_row_offsets = maz_counts[taz_choices_trace_targets].cumsum()
trace_inserts = np.repeat(trace_last_row_offsets,
max_maz_count - trace_maz_counts)
# trace dest_maz_alts
padded_maz_sizes = np.insert(trace_maz_sizes[CHOOSER_ID].values,
trace_inserts,
0.0).reshape(-1, max_maz_count)
df = pd.DataFrame(data=padded_maz_sizes,
columns=alt_dest_columns,
index=trace_taz_choices_df.index)
df = pd.concat([lhs_df, df], axis=1)
tracing.trace_df(df,
label=tracing.extend_trace_label(
trace_label, 'dest_maz_alts'),
transpose=False)
# trace dest_maz_size_terms
padded_maz_sizes = np.insert(trace_maz_sizes['size_term'].values,
trace_inserts,
0.0).reshape(-1, max_maz_count)
df = pd.DataFrame(data=padded_maz_sizes,
columns=alt_dest_columns,
index=trace_taz_choices_df.index)
df = pd.concat([lhs_df, df], axis=1)
tracing.trace_df(df,
label=tracing.extend_trace_label(
trace_label, 'dest_maz_size_terms'),
transpose=False)
# trace dest_maz_probs
df = pd.DataFrame(data=maz_probs[taz_choices_trace_targets],
columns=alt_dest_columns,
index=trace_taz_choices_df.index)
df = pd.concat([lhs_df, df], axis=1)
df['rand'] = rands[taz_choices_trace_targets]
tracing.trace_df(df,
label=tracing.extend_trace_label(
trace_label, 'dest_maz_probs'),
transpose=False)
taz_choices = taz_choices.drop(columns=['TAZ_prob', 'MAZ_prob'])
taz_choices = taz_choices.groupby([chooser_id_col, DEST_MAZ
]).agg(prob=('prob', 'max'),
pick_count=('prob', 'count'))
taz_choices.reset_index(level=DEST_MAZ, inplace=True)
return taz_choices
def compute_columns(df, model_settings, locals_dict={}, trace_label=None):
"""
Evaluate expressions_spec in context of df, with optional additional pipeline tables in locals
Parameters
----------
df : pandas DataFrame
or if None, expect name of pipeline table to be specified by DF in model_settings
model_settings : dict or str
dict with keys:
DF - df_alias and (additionally, if df is None) name of pipeline table to load as df
SPEC - name of expressions file (csv suffix optional) if different from model_settings
TABLES - list of pipeline tables to load and make available as (read only) locals
str:
name of yaml file in configs_dir to load dict from
locals_dict : dict
dict of locals (e.g. utility functions) to add to the execution environment
trace_label
Returns
-------
results: pandas.DataFrame
one column for each expression (except temps with ALL_CAP target names)
same index as df
"""
if isinstance(model_settings, str):
model_settings_name = model_settings
model_settings = config.read_model_settings('%s.yaml' % model_settings)
assert model_settings, "Found no model settings for %s" % model_settings_name
else:
model_settings_name = 'dict'
assert isinstance(model_settings, dict)
assert 'DF' in model_settings, \
"Expected to find 'DF' in %s" % model_settings_name
df_name = model_settings.get('DF')
helper_table_names = model_settings.get('TABLES', [])
expressions_spec_name = model_settings.get('SPEC', None)
assert expressions_spec_name is not None, \
"Expected to find 'SPEC' in %s" % model_settings_name
trace_label = tracing.extend_trace_label(trace_label or '',
expressions_spec_name)
if not expressions_spec_name.endswith(".csv"):
expressions_spec_name = '%s.csv' % expressions_spec_name
logger.debug(
f"{trace_label} compute_columns using expression spec file {expressions_spec_name}"
)
expressions_spec = assign.read_assignment_spec(
config.config_file_path(expressions_spec_name))
assert expressions_spec.shape[0] > 0, \
"Expected to find some assignment expressions in %s" % expressions_spec_name
tables = {t: inject.get_table(t).to_frame() for t in helper_table_names}
# if df was passed in, df might be a slice, or any other table, but DF is it's local alias
assert df_name not in tables, "Did not expect to find df '%s' in TABLES" % df_name
tables[df_name] = df
# be nice and also give it to them as df?
tables['df'] = df
_locals_dict = assign.local_utilities()
_locals_dict.update(locals_dict)
_locals_dict.update(tables)
# FIXME a number of asim model preprocessors want skim_dict - should they request it in model_settings.TABLES?
_locals_dict.update({
# 'los': inject.get_injectable('network_los', None),
'skim_dict': inject.get_injectable('skim_dict', None),
})
results, trace_results, trace_assigned_locals \
= assign.assign_variables(expressions_spec,
df,
_locals_dict,
trace_rows=tracing.trace_targets(df))
if trace_results is not None:
tracing.trace_df(trace_results, label=trace_label, slicer='NONE')
if trace_assigned_locals:
tracing.write_csv(trace_assigned_locals,
file_name="%s_locals" % trace_label)
return results
