def write_csv(df,
file_name,
index_label=None,
columns=None,
column_labels=None,
transpose=True):
"""
Print write_csv
Parameters
----------
df: pandas.DataFrame or pandas.Series
traced dataframe
file_name: str
output file name
index_label: str
index name
columns: list
columns to write
transpose: bool
whether to transpose dataframe (ignored for series)
Returns
-------
Nothing
"""
assert len(file_name) > 0
if not file_name.endswith('.%s' % CSV_FILE_TYPE):
file_name = '%s.%s' % (file_name, CSV_FILE_TYPE)
file_path = config.trace_file_path(file_name)
if os.path.isfile(file_path):
logger.debug("write_csv file exists %s %s" %
(type(df).__name__, file_name))
if isinstance(df, pd.DataFrame):
# logger.debug("dumping %s dataframe to %s" % (df.shape, file_name))
write_df_csv(df,
file_path,
index_label,
columns,
column_labels,
transpose=transpose)
elif isinstance(df, pd.Series):
# logger.debug("dumping %s element series to %s" % (df.shape[0], file_name))
write_series_csv(df, file_path, index_label, columns, column_labels)
elif isinstance(df, dict):
df = pd.Series(data=df)
# logger.debug("dumping %s element dict to %s" % (df.shape[0], file_name))
write_series_csv(df, file_path, index_label, columns, column_labels)
else:
logger.error(
"write_csv object for file_name '%s' of unexpected type: %s" %
(file_name, type(df)))
def trace_interaction_eval_results(trace_results, trace_ids, label):
"""
Trace model design eval results for interaction_simulate
Parameters
----------
trace_results: pandas.DataFrame
traced model_design dataframe
trace_ids : tuple (str, numpy.ndarray)
column name and array of trace_ids from interaction_trace_rows()
used to filter the trace_results dataframe by traced hh or person id
label: str
tracer name
Returns
-------
Nothing
"""
assert type(trace_ids[1]) == np.ndarray
slicer_column_name = trace_ids[0]
trace_results[slicer_column_name] = trace_ids[1]
targets = np.unique(trace_ids[1])
if len(trace_results.index) == 0:
return
# write out the raw dataframe
file_path = config.trace_file_path('%s.raw.csv' % label)
trace_results.to_csv(file_path, mode="a", index=True, header=True)
# if there are multiple targets, we want them in separate tables for readability
for target in targets:
df_target = trace_results[trace_results[slicer_column_name] == target]
# we want the transposed columns in predictable order
df_target.sort_index(inplace=True)
# # remove the slicer (person_id or hh_id) column?
# del df_target[slicer_column_name]
target_label = '%s.%s.%s' % (label, slicer_column_name, target)
trace_df(df_target,
label=target_label,
slicer="NONE",
transpose=True,
column_labels=['expression', None],
warn_if_empty=False)
def trace_interaction_eval_results(trace_results, trace_ids, label):
"""
Trace model design eval results for interaction_simulate
Parameters
----------
trace_results: pandas.DataFrame
traced model_design dataframe
trace_ids : tuple (str, numpy.ndarray)
column name and array of trace_ids from interaction_trace_rows()
used to filter the trace_results dataframe by traced hh or person id
label: str
tracer name
Returns
-------
Nothing
"""
assert type(trace_ids[1]) == np.ndarray
slicer_column_name = trace_ids[0]
trace_results[slicer_column_name] = trace_ids[1]
targets = np.unique(trace_ids[1])
if len(trace_results.index) == 0:
return
# write out the raw dataframe
file_path = config.trace_file_path('%s.raw.csv' % label)
trace_results.to_csv(file_path, mode="a", index=True, header=True)
# if there are multiple targets, we want them in separate tables for readability
for target in targets:
df_target = trace_results[trace_results[slicer_column_name] == target]
# we want the transposed columns in predictable order
df_target.sort_index(inplace=True)
# # remove the slicer (person_id or hh_id) column?
# del df_target[slicer_column_name]
target_label = '%s.%s.%s' % (label, slicer_column_name, target)
trace_df(df_target,
label=target_label,
slicer="NONE",
transpose=True,
column_labels=['expression', None],
warn_if_empty=False)
def write_csv(df, file_name, index_label=None, columns=None, column_labels=None, transpose=True):
"""
Print write_csv
Parameters
----------
df: pandas.DataFrame or pandas.Series
traced dataframe
file_name: str
output file name
index_label: str
index name
columns: list
columns to write
transpose: bool
whether to transpose dataframe (ignored for series)
Returns
-------
Nothing
"""
assert len(file_name) > 0
if not file_name.endswith('.%s' % CSV_FILE_TYPE):
file_name = '%s.%s' % (file_name, CSV_FILE_TYPE)
file_path = config.trace_file_path(file_name)
if os.path.isfile(file_path):
logger.debug("write_csv file exists %s %s" % (type(df).__name__, file_name))
if isinstance(df, pd.DataFrame):
# logger.debug("dumping %s dataframe to %s" % (df.shape, file_name))
write_df_csv(df, file_path, index_label, columns, column_labels, transpose=transpose)
elif isinstance(df, pd.Series):
# logger.debug("dumping %s element series to %s" % (df.shape[0], file_name))
write_series_csv(df, file_path, index_label, columns, column_labels)
elif isinstance(df, dict):
df = pd.Series(data=df)
# logger.debug("dumping %s element dict to %s" % (df.shape[0], file_name))
write_series_csv(df, file_path, index_label, columns, column_labels)
else:
logger.error("write_csv object for file_name '%s' of unexpected type: %s" %
(file_name, type(df)))