def test_percentiles(self):
with warnings.catch_warnings():
warnings.simplefilter('error')
with self.assertRaises(NullCalculationWarning):
Percentiles('one').validate(self.table)
with self.assertRaises(DataTypeError):
Percentiles('three').validate(self.table)
warnings.simplefilter('ignore')
try:
Percentiles('two').validate(self.table)
finally:
warnings.resetwarnings()
rows = [(n, ) for n in range(1, 1001)]
table = Table(rows, ['ints'], [self.number_type])
percentiles = Percentiles('ints').run(table)
self.assertEqual(percentiles[0], Decimal('1'))
self.assertEqual(percentiles[25], Decimal('250.5'))
self.assertEqual(percentiles[50], Decimal('500.5'))
self.assertEqual(percentiles[75], Decimal('750.5'))
self.assertEqual(percentiles[99], Decimal('990.5'))
self.assertEqual(percentiles[100], Decimal('1000'))
def prepare(self, table):
column = table.columns[self._column_name]
if not isinstance(column.data_type, Number):
raise DataTypeError(
'PercentileRank column must contain Number data.')
self._percentiles = Percentiles(self._column_name).run(table)
def run(self, table):
"""
:returns:
:class:`int`
"""
percentiles = Percentiles(self._column_name).run(table)
new_column = []
for row in table.rows:
new_column.append(percentiles.locate(row[self._column_name]))
return new_column
def run(self, table):
"""
:returns:
:class:`int`
"""
percentiles = Percentiles(self._column_name).run(table)
new_column = []
for row in table.rows:
new_column.append(percentiles.locate(row[self._column_name]))
return new_column
def prepare(self, table):
column = table.columns[self._column_name]
if not isinstance(column.data_type, Number):
raise DataTypeError('PercentileRank column must contain Number data.')
self._percentiles = Percentiles(self._column_name).run(table)
def test_percentiles_locate(self):
rows = [(n, ) for n in range(1, 1001)]
table = Table(rows, ['ints'], [self.number_type])
percentiles = Percentiles('ints').run(table)
self.assertEqual(percentiles.locate(251), Decimal('25'))
self.assertEqual(percentiles.locate(260), Decimal('25'))
self.assertEqual(percentiles.locate(261), Decimal('26'))
with self.assertRaises(ValueError):
percentiles.locate(0)
with self.assertRaises(ValueError):
percentiles.locate(1012)
class PercentileRank(Rank):
"""
Assign each value in a column to the percentile into which it falls.
See :class:`.Percentiles` for implementation details.
"""
def prepare(self, table):
column = table.columns[self._column_name]
if not isinstance(column.data_type, Number):
raise DataTypeError('PercentileRank column must contain Number data.')
self._percentiles = Percentiles(self._column_name).run(table)
def run(self, row):
"""
:returns:
:class:`int`
"""
return self._percentiles.locate(row[self._column_name])
class PercentileRank(Rank):
"""
Assign each value in a column to the percentile into which it falls.
See :class:`.Percentiles` for implementation details.
"""
def prepare(self, table):
column = table.columns[self._column_name]
if not isinstance(column.data_type, Number):
raise DataTypeError(
'PercentileRank column must contain Number data.')
self._percentiles = Percentiles(self._column_name).run(table)
def run(self, row):
"""
:returns:
:class:`int`
"""
return self._percentiles.locate(row[self._column_name])
def test_percentiles_all_nulls(self):
self.assertEqual(
Percentiles('four').run(self.table), Quantiles([None] * 101))
