def trade_flows(self, from_country=None, to_country=None, sector=None, with_RoW=False):
"""
Per-sector country-country flows.
Calculated from the global import vector and the
import propensities. Use `from_country`, `to_country` and `sector` to filter
"""
imports = self.imports
imports = dataframe.filter_pandas(imports, 'sector', sector)
imports = dataframe.filter_pandas(imports, 'to_country', to_country)
flows = self._split_flows_by_import_propensities(
imports, country_field='to_country', with_RoW=with_RoW)
flows = dataframe.filter_pandas(flows, 'from_country', from_country)
if not with_RoW:
flows = dataframe.filter_pandas(flows, ['from_country', 'to_country'], 'RoW', exclude=True)
flows.name = 'value'
return flows
def _filter_flows(self, flows, country_names=None, sectors=None,
with_RoW=False):
"""
Filter a MultiIndexed pandas object by `country_names` and
`sectors`.
Both `country_names` and `sectors` can be None, a string or a list of
strings.
All of `to_country`, `from_country` and `country` are
filtered.
"""
if country_names is None and sectors is None and with_RoW:
return flows
else:
# What should we filter the pandas object on? (index or columns?)
if flows.index.names[0] is None:
# Dataframe/Series is unindexed
names = flows.columns.values.tolist()
else:
names = flows.index.names
# Prepare the filtered pandas object
filtered = flows.copy()
if country_names is None:
country_names = self.country_names
# Remove RoW
if with_RoW:
country_names = country_names + ['RoW']
if sectors is None:
sectors = self.sectors
# Level names on which to filter
sector_levels = [l for l in names if str(l).find('sector') >= 0]
country_levels = [l for l in names if str(l).find('country') >= 0]
for level in sector_levels:
filtered = dataframe.filter_pandas(filtered, level, sectors)
for level in country_levels:
filtered = dataframe.filter_pandas(filtered, level, country_names)
return filtered
