def test_custom_primitive_time_as_arg(es):
def time_since_last(values, time):
time_since = time - values.iloc[0]
return time_since.total_seconds()
TimeSinceLast = make_agg_primitive(time_since_last,
[DatetimeTimeIndex],
Numeric,
uses_calc_time=True)
assert TimeSinceLast.name == "time_since_last"
f = TimeSinceLast(es["log"]["datetime"], es["customers"])
fm = ft.calculate_feature_matrix([f],
entityset=es,
instance_ids=[0, 1, 2],
cutoff_time=datetime(2015, 6, 8))
correct = [131376600, 131289600, 131287800]
# note: must round to nearest second
assert all(fm[f.get_name()].round().values == correct)
error_text = "'time' is a restricted keyword. Please use a different keyword."
with pytest.raises(ValueError, match=error_text):
make_agg_primitive(time_since_last,
[DatetimeTimeIndex],
Numeric,
uses_calc_time=False)
def test_time_since_last_custom(pd_es):
def time_since_last(values, time=None):
time_since = time - values.iloc[0]
return time_since.total_seconds()
TimeSinceLast = make_agg_primitive(
time_since_last,
[ColumnSchema(logical_type=Datetime, semantic_tags={'time_index'})],
ColumnSchema(semantic_tags={'numeric'}),
name="time_since_last",
uses_calc_time=True)
f = ft.Feature(pd_es["log"].ww["datetime"],
parent_dataframe_name="customers",
primitive=TimeSinceLast)
fm = ft.calculate_feature_matrix([f],
entityset=pd_es,
instance_ids=[0, 1, 2],
cutoff_time=datetime(2015, 6, 8))
correct = [131376600, 131289600, 131287800]
# note: must round to nearest second
assert all(fm[f.get_name()].round().values == correct)
error_text = "'time' is a restricted keyword. Please use a different keyword."
with pytest.raises(ValueError, match=error_text):
TimeSinceLast = make_agg_primitive(time_since_last, [
ColumnSchema(logical_type=Datetime, semantic_tags={'time_index'})
],
ColumnSchema(
semantic_tags={'numeric'}),
uses_calc_time=False)
def test_custom_primitive_time_as_arg(es):
def time_since_last(values, time):
time_since = time - values.iloc[0]
return time_since.total_seconds()
TimeSinceLast = make_agg_primitive(time_since_last, [DatetimeTimeIndex],
Numeric,
uses_calc_time=True)
assert TimeSinceLast.name == "time_since_last"
f = ft.Feature(es["log"]["datetime"],
parent_entity=es["customers"],
primitive=TimeSinceLast)
fm = ft.calculate_feature_matrix([f],
entityset=es,
instance_ids=[0, 1, 2],
cutoff_time=datetime(2015, 6, 8))
correct = [131376600, 131289600, 131287800]
# note: must round to nearest second
assert all(fm[f.get_name()].round().values == correct)
error_text = "'time' is a restricted keyword. Please use a different keyword."
with pytest.raises(ValueError, match=error_text):
make_agg_primitive(time_since_last, [DatetimeTimeIndex],
Numeric,
uses_calc_time=False)
def test_agg_same_method_name(es):
"""
Pandas relies on the function name when calculating aggregations. This means if a two
primitives with the same function name are applied to the same column, pandas
can't differentiate them. We have a work around to this based on the name property
that we test here.
"""
# TODO: Update to work with Dask and Koalas
if not all(isinstance(entity.df, pd.DataFrame) for entity in es.entities):
pytest.xfail(
"Cannot use primitives made with make_agg_primitives with Dask or Koalas EntitySets"
)
# test with normally defined functions
def custom_primitive(x):
return x.sum()
Sum = make_agg_primitive(custom_primitive,
input_types=[Numeric],
return_type=Numeric,
name="sum")
def custom_primitive(x):
return x.max()
Max = make_agg_primitive(custom_primitive,
input_types=[Numeric],
return_type=Numeric,
name="max")
f_sum = ft.Feature(es["log"]["value"],
parent_entity=es["customers"],
primitive=Sum)
f_max = ft.Feature(es["log"]["value"],
parent_entity=es["customers"],
primitive=Max)
fm = ft.calculate_feature_matrix([f_sum, f_max], entityset=es)
assert fm.columns.tolist() == [f_sum.get_name(), f_max.get_name()]
# test with lambdas
Sum = make_agg_primitive(lambda x: x.sum(),
input_types=[Numeric],
return_type=Numeric,
name="sum")
Max = make_agg_primitive(lambda x: x.max(),
input_types=[Numeric],
return_type=Numeric,
name="max")
f_sum = ft.Feature(es["log"]["value"],
parent_entity=es["customers"],
primitive=Sum)
f_max = ft.Feature(es["log"]["value"],
parent_entity=es["customers"],
primitive=Max)
fm = ft.calculate_feature_matrix([f_sum, f_max], entityset=es)
assert fm.columns.tolist() == [f_sum.get_name(), f_max.get_name()]
def test_agg_same_method_name(es):
"""
Pandas relies on the function name when calculating aggregations. This means if a two
primitives with the same function name are applied to the same column, pandas
can't differentiate them. We have a work around to this based on the name property
that we test here.
"""
# test with normally defined functions
def custom_primitive(x):
return x.sum()
Sum = make_agg_primitive(custom_primitive,
input_types=[Numeric],
return_type=Numeric,
name="sum")
def custom_primitive(x):
return x.max()
Max = make_agg_primitive(custom_primitive,
input_types=[Numeric],
return_type=Numeric,
name="max")
f_sum = ft.Feature(es["log"]["value"],
parent_entity=es["customers"],
primitive=Sum)
f_max = ft.Feature(es["log"]["value"],
parent_entity=es["customers"],
primitive=Max)
fm = ft.calculate_feature_matrix([f_sum, f_max], entityset=es)
assert fm.columns.tolist() == [f_sum.get_name(), f_max.get_name()]
# test with lambdas
Sum = make_agg_primitive(lambda x: x.sum(),
input_types=[Numeric],
return_type=Numeric,
name="sum")
Max = make_agg_primitive(lambda x: x.max(),
input_types=[Numeric],
return_type=Numeric,
name="max")
f_sum = ft.Feature(es["log"]["value"],
parent_entity=es["customers"],
primitive=Sum)
f_max = ft.Feature(es["log"]["value"],
parent_entity=es["customers"],
primitive=Max)
fm = ft.calculate_feature_matrix([f_sum, f_max], entityset=es)
assert fm.columns.tolist() == [f_sum.get_name(), f_max.get_name()]
def test_make_three_most_common(pd_es):
def pd_top3(x):
array = np.array(x.value_counts()[:3].index)
if len(array) < 3:
filler = np.full(3 - len(array), np.nan)
array = np.append(array, filler)
return array
NMostCommoner = make_agg_primitive(function=pd_top3,
input_types=[Discrete],
return_type=Discrete,
number_output_features=3)
fm, features = ft.dfs(entityset=pd_es,
target_entity="customers",
instance_ids=[0, 1, 2],
agg_primitives=[NMostCommoner],
trans_primitives=[])
df = fm[["PD_TOP3(log.product_id)[%s]" % i for i in range(3)]]
assert set(df.iloc[0].values[:2]) == set([
'coke zero', 'toothpaste'
]) # coke zero and toothpaste have same number of occurrences
assert df.iloc[0].values[2] in ['car', 'brown bag'
] # so just check that the top two match
assert df.iloc[1].reset_index(drop=True).equals(
pd.Series(['coke zero', 'Haribo sugar-free gummy bears', np.nan]))
assert df.iloc[2].reset_index(drop=True).equals(
pd.Series(['taco clock', np.nan, np.nan]))
def test_override_multi_feature_names(pd_es):
def gen_custom_names(primitive, base_feature_names, relationship_path_name,
parent_entity_id, where_str, use_prev_str):
base_string = 'Custom_%s({}.{})'.format(parent_entity_id, base_feature_names)
return [base_string % i for i in range(primitive.number_output_features)]
def pd_top3(x):
array = np.array(x.value_counts()[:3].index)
if len(array) < 3:
filler = np.full(3 - len(array), np.nan)
array = np.append(array, filler)
return array
num_features = 3
NMostCommoner = make_agg_primitive(function=pd_top3,
input_types=[Numeric],
return_type=Discrete,
number_output_features=num_features,
cls_attributes={"generate_names": gen_custom_names})
fm, features = ft.dfs(entityset=pd_es,
target_entity="products",
instance_ids=[0, 1, 2],
agg_primitives=[NMostCommoner],
trans_primitives=[])
expected_names = []
base_names = [['value'], ['value_2'], ['value_many_nans']]
for name in base_names:
expected_names += gen_custom_names(NMostCommoner, name, None, 'products', None, None)
for name in expected_names:
assert name in fm.columns
def test_pickle_features_with_custom_primitive(es):
NewMean = make_agg_primitive(
np.nanmean,
name="NewMean",
input_types=[Numeric],
return_type=Numeric,
description="Calculate means ignoring nan values")
dfs_obj = DeepFeatureSynthesis(target_entity_id='sessions',
entityset=es,
agg_primitives=[Last, Mean, NewMean],
trans_primitives=[],
max_features=20)
features_no_pickle = dfs_obj.build_features()
assert any([isinstance(feat, NewMean) for feat in features_no_pickle])
dir_path = os.path.dirname(os.path.realpath(__file__))
filepath = os.path.join(dir_path, 'test_feature')
es_filepath = os.path.join(dir_path, 'test_entityset')
# pickle entityset
save_obj_pickle(es, es_filepath)
ft.save_features(features_no_pickle, filepath)
features_pickle = ft.load_features(filepath)
for feat_1, feat_2 in zip(features_no_pickle, features_pickle):
assert feat_1.hash() == feat_2.hash()
assert feat_1.entityset == feat_2.entityset
# file is smaller than entityset in memory
assert os.path.getsize(filepath) < asizeof(es)
# file is smaller than entityset pickled
assert os.path.getsize(filepath) < os.path.getsize(es_filepath)
os.remove(filepath)
os.remove(es_filepath)
def test_custom_primitive_multiple_inputs(es):
def mean_sunday(numeric, datetime):
'''
Finds the mean of non-null values of a feature that occurred on Sundays
'''
days = pd.DatetimeIndex(datetime).weekday.values
df = pd.DataFrame({'numeric': numeric, 'time': days})
return df[df['time'] == 6]['numeric'].mean()
MeanSunday = make_agg_primitive(function=mean_sunday,
input_types=[Numeric, Datetime],
return_type=Numeric)
fm, features = ft.dfs(entityset=es,
target_entity="sessions",
agg_primitives=[MeanSunday],
trans_primitives=[])
mean_sunday_value = pd.Series([None, None, None, 2.5, 7, None])
iterator = zip(fm["MEAN_SUNDAY(log.value, datetime)"], mean_sunday_value)
for x, y in iterator:
assert ((pd.isnull(x) and pd.isnull(y)) or (x == y))
es.add_interesting_values()
mean_sunday_value_priority_0 = pd.Series([None, None, None, 2.5, 0, None])
fm, features = ft.dfs(entityset=es,
target_entity="sessions",
agg_primitives=[MeanSunday],
trans_primitives=[],
where_primitives=[MeanSunday])
where_feat = "MEAN_SUNDAY(log.value, datetime WHERE priority_level = 0)"
for x, y in zip(fm[where_feat], mean_sunday_value_priority_0):
assert ((pd.isnull(x) and pd.isnull(y)) or (x == y))
def test_count_null_and_make_agg_primitive(pd_es):
def count_func(values, count_null=False):
if len(values) == 0:
return 0
if count_null:
values = values.fillna(0)
return values.count()
def count_generate_name(self, base_feature_names, relationship_path_name,
parent_dataframe_name, where_str, use_prev_str):
return u"COUNT(%s%s%s)" % (relationship_path_name, where_str,
use_prev_str)
Count = make_agg_primitive(
count_func,
[[ColumnSchema(semantic_tags={'foreign_key'})], [ColumnSchema()]],
ColumnSchema(semantic_tags={'numeric'}),
name="count",
stack_on_self=False,
cls_attributes={"generate_name": count_generate_name})
count_null = ft.Feature(pd_es['log'].ww['value'],
parent_dataframe_name='sessions',
primitive=Count(count_null=True))
feature_matrix = ft.calculate_feature_matrix([count_null], entityset=pd_es)
values = [5, 4, 1, 2, 3, 2]
assert (values == feature_matrix[count_null.get_name()]).all()
def test_custom_primitive_multiple_inputs(es):
def mean_sunday(numeric, datetime):
'''
Finds the mean of non-null values of a feature that occurred on Sundays
'''
days = pd.DatetimeIndex(datetime).weekday.values
df = pd.DataFrame({'numeric': numeric, 'time': days})
return df[df['time'] == 6]['numeric'].mean()
MeanSunday = make_agg_primitive(function=mean_sunday,
input_types=[Numeric, Datetime],
return_type=Numeric)
fm, features = ft.dfs(entityset=es,
target_entity="sessions",
agg_primitives=[MeanSunday],
trans_primitives=[])
mean_sunday_value = pd.Series([None, None, None, 2.5, 7, None])
iterator = zip(fm["MEAN_SUNDAY(log.value, datetime)"], mean_sunday_value)
for x, y in iterator:
assert ((pd.isnull(x) and pd.isnull(y)) or (x == y))
es.add_interesting_values()
mean_sunday_value_priority_0 = pd.Series([None, None, None, 2.5, 0, None])
fm, features = ft.dfs(entityset=es,
target_entity="sessions",
agg_primitives=[MeanSunday],
trans_primitives=[],
where_primitives=[MeanSunday])
where_feat = "MEAN_SUNDAY(log.value, datetime WHERE priority_level = 0)"
for x, y in zip(fm[where_feat], mean_sunday_value_priority_0):
assert ((pd.isnull(x) and pd.isnull(y)) or (x == y))
def test_make_three_most_common(es):
def pd_top3(x):
array = np.array(x.value_counts()[:3].index)
if len(array) < 3:
filler = np.full(3 - len(array), np.nan)
array = np.append(array, filler)
return array
NMostCommoner = make_agg_primitive(function=pd_top3,
input_types=[Discrete],
return_type=Discrete,
number_output_features=3)
fm, features = ft.dfs(entityset=es,
target_entity="customers",
agg_primitives=[NMostCommoner],
trans_primitives=[])
true_results = pd.DataFrame(
[['coke zero', 'toothpaste', "car"],
['coke zero', 'Haribo sugar-free gummy bears', np.nan],
['taco clock', np.nan, np.nan]])
df = fm[["PD_TOP3(log.product_id)__%s" % i for i in range(3)]]
for i in range(df.shape[0]):
if i == 0:
# coke zero and toothpaste have same number of occurrences
# so just check that the top two match
assert set(true_results.iloc[i].values[:2]) == set(
df.iloc[i].values[:2])
assert df.iloc[0].values[2] in ("brown bag", "car")
else:
for i1, i2 in zip(true_results.iloc[i], df.iloc[i]):
assert (pd.isnull(i1) and pd.isnull(i2)) or (i1 == i2)
def test_count_null_and_make_agg_primitive(es):
def count_func(values, count_null=False):
if len(values) == 0:
return 0
if count_null:
values = values.fillna(0)
return values.count()
def count_generate_name(self, base_feature_names, relationship_path_name,
parent_entity_id, where_str, use_prev_str):
return u"COUNT(%s%s%s)" % (relationship_path_name, where_str,
use_prev_str)
Count = make_agg_primitive(
count_func, [[Index], [Variable]],
Numeric,
name="count",
stack_on_self=False,
cls_attributes={"generate_name": count_generate_name})
count_null = ft.Feature(es['log']['value'],
parent_entity=es['sessions'],
primitive=Count(count_null=True))
feature_matrix = ft.calculate_feature_matrix([count_null], entityset=es)
values = [5, 4, 1, 2, 3, 2]
assert (values == feature_matrix[count_null.get_name()]).all()
def test_uses_calc_time():
def time_since_last(values, time=None):
time_since = time - values.iloc[0]
return time_since.total_seconds()
TimeSinceLast = make_agg_primitive(time_since_last, [DatetimeTimeIndex],
Numeric,
name="time_since_last",
uses_calc_time=True)
primitive = TimeSinceLast()
datetimes = pd.Series([datetime(2015, 6, 7), datetime(2015, 6, 6)])
answer = 86400.0
assert answer == primitive(datetimes, time=datetime(2015, 6, 8))
def test_agg_same_method_name(es):
"""
Pandas relies on the function name when calculating aggregations. This means if a two
primitives with the same function name are applied to the same column, pandas
can't differentiate them. We have a work around to this based on the name property
that we test here.
"""
# test with normally defined functions
def custom_primitive(x):
return x.sum()
Sum = make_agg_primitive(custom_primitive, input_types=[Numeric],
return_type=Numeric, name="sum")
def custom_primitive(x):
return x.max()
Max = make_agg_primitive(custom_primitive, input_types=[Numeric],
return_type=Numeric, name="max")
f_sum = Sum(es["log"]["value"], es["customers"])
f_max = Max(es["log"]["value"], es["customers"])
fm = ft.calculate_feature_matrix([f_sum, f_max], entityset=es)
assert fm.columns.tolist() == [f_sum.get_name(), f_max.get_name()]
# test with lambdas
Sum = make_agg_primitive(lambda x: x.sum(), input_types=[Numeric],
return_type=Numeric, name="sum")
Max = make_agg_primitive(lambda x: x.max(), input_types=[Numeric],
return_type=Numeric, name="max")
f_sum = Sum(es["log"]["value"], es["customers"])
f_max = Max(es["log"]["value"], es["customers"])
fm = ft.calculate_feature_matrix([f_sum, f_max], entityset=es)
assert fm.columns.tolist() == [f_sum.get_name(), f_max.get_name()]
def test_custom_primitive_default_kwargs(es):
def sum_n_times(numeric, n=1):
return np.nan_to_num(numeric).sum(dtype=np.float) * n
SumNTimes = make_agg_primitive(function=sum_n_times,
input_types=[Numeric],
return_type=Numeric)
sum_n_1_n = 1
sum_n_1_base_f = ft.Feature(es['log']['value'])
sum_n_1 = ft.Feature([sum_n_1_base_f], parent_entity=es['sessions'], primitive=SumNTimes(n=sum_n_1_n))
sum_n_2_n = 2
sum_n_2_base_f = ft.Feature(es['log']['value_2'])
sum_n_2 = ft.Feature([sum_n_2_base_f], parent_entity=es['sessions'], primitive=SumNTimes(n=sum_n_2_n))
assert sum_n_1_base_f == sum_n_1.base_features[0]
assert sum_n_1_n == sum_n_1.primitive.kwargs['n']
assert sum_n_2_base_f == sum_n_2.base_features[0]
assert sum_n_2_n == sum_n_2.primitive.kwargs['n']
def test_custom_primitive_default_kwargs(es):
def sum_n_times(numeric, n=1):
return np.nan_to_num(numeric).sum(dtype=np.float) * n
SumNTimes = make_agg_primitive(function=sum_n_times,
input_types=[Numeric],
return_type=Numeric)
sum_n_1_n = 1
sum_n_1_base_f = Feature(es['log']['value'])
sum_n_1 = SumNTimes([sum_n_1_base_f], es['sessions'], n=sum_n_1_n)
sum_n_2_n = 2
sum_n_2_base_f = Feature(es['log']['value_2'])
sum_n_2 = SumNTimes([sum_n_2_base_f], es['sessions'], n=sum_n_2_n)
assert sum_n_1_base_f == sum_n_1.base_features[0]
assert sum_n_1_n == sum_n_1.kwargs['n']
assert sum_n_2_base_f == sum_n_2.base_features[0]
assert sum_n_2_n == sum_n_2.kwargs['n']
def test_custom_primitive_default_kwargs(es):
def sum_n_times(numeric, n=1):
return np.nan_to_num(numeric).sum(dtype=np.float) * n
SumNTimes = make_agg_primitive(
function=sum_n_times,
input_types=[ColumnSchema(semantic_tags={'numeric'})],
return_type=ColumnSchema(semantic_tags={'numeric'}))
sum_n_1_n = 1
sum_n_1_base_f = ft.Feature(es['log'].ww['value'])
sum_n_1 = ft.Feature([sum_n_1_base_f],
parent_dataframe_name='sessions',
primitive=SumNTimes(n=sum_n_1_n))
sum_n_2_n = 2
sum_n_2_base_f = ft.Feature(es['log'].ww['value_2'])
sum_n_2 = ft.Feature([sum_n_2_base_f],
parent_dataframe_name='sessions',
primitive=SumNTimes(n=sum_n_2_n))
assert sum_n_1_base_f == sum_n_1.base_features[0]
assert sum_n_1_n == sum_n_1.primitive.kwargs['n']
assert sum_n_2_base_f == sum_n_2.base_features[0]
assert sum_n_2_n == sum_n_2.primitive.kwargs['n']
def test_count_null_and_make_agg_primitive(es):
def count_func(values, count_null=False):
if len(values) == 0:
return 0
if count_null:
values = values.fillna(0)
return values.count()
def count_generate_name(self):
where_str = self._where_str()
use_prev_str = self._use_prev_str()
return u"COUNT(%s%s%s)" % (self.child_entity.id,
where_str,
use_prev_str)
Count = make_agg_primitive(count_func, [[Index], [Variable]], Numeric,
name="count", stack_on_self=False,
cls_attributes={"generate_name": count_generate_name})
count_null = Count(es['log']['value'], es['sessions'], count_null=True)
feature_matrix = ft.calculate_feature_matrix([count_null], entityset=es)
values = [5, 4, 1, 2, 3, 2]
assert (values == feature_matrix[count_null.get_name()]).all()
def test_agg_same_method_name(es):
"""
Pandas relies on the function name when calculating aggregations. This means if a two
primitives with the same function name are applied to the same column, pandas
can't differentiate them. We have a work around to this based on the name property
that we test here.
"""
# TODO: Update to work with Dask and Koalas
if es.dataframe_type != Library.PANDAS.value:
pytest.xfail(
"Cannot use primitives made with make_agg_primitives with Dask or Koalas EntitySets"
)
# test with normally defined functions
def custom_primitive(x):
return x.sum()
Sum = make_agg_primitive(
custom_primitive,
input_types=[ColumnSchema(semantic_tags={'numeric'})],
return_type=ColumnSchema(semantic_tags={'numeric'}),
name="sum")
def custom_primitive(x):
return x.max()
Max = make_agg_primitive(
custom_primitive,
input_types=[ColumnSchema(semantic_tags={'numeric'})],
return_type=ColumnSchema(semantic_tags={'numeric'}),
name="max")
f_sum = ft.Feature(es["log"].ww["value"],
parent_dataframe_name="customers",
primitive=Sum)
f_max = ft.Feature(es["log"].ww["value"],
parent_dataframe_name="customers",
primitive=Max)
fm = ft.calculate_feature_matrix([f_sum, f_max], entityset=es)
assert fm.columns.tolist() == [f_sum.get_name(), f_max.get_name()]
# test with lambdas
Sum = make_agg_primitive(
lambda x: x.sum(),
input_types=[ColumnSchema(semantic_tags={'numeric'})],
return_type=ColumnSchema(semantic_tags={'numeric'}),
name="sum")
Max = make_agg_primitive(
lambda x: x.max(),
input_types=[ColumnSchema(semantic_tags={'numeric'})],
return_type=ColumnSchema(semantic_tags={'numeric'}),
name="max")
f_sum = ft.Feature(es["log"].ww["value"],
parent_dataframe_name="customers",
primitive=Sum)
f_max = ft.Feature(es["log"].ww["value"],
parent_dataframe_name="customers",
primitive=Max)
fm = ft.calculate_feature_matrix([f_sum, f_max], entityset=es)
assert fm.columns.tolist() == [f_sum.get_name(), f_max.get_name()]
from woodwork.column_schema import ColumnSchema
from featuretools.primitives.base import make_agg_primitive
CustomSum = make_agg_primitive(
lambda x: sum(x),
name="CustomSum",
input_types=[ColumnSchema(semantic_tags={'numeric'})],
return_type=ColumnSchema(semantic_tags={'numeric'}))
from featuretools.primitives.base import make_agg_primitive
from featuretools.variable_types import Numeric
CustomMean = make_agg_primitive(lambda x: sum(x) / len(x),
name="CustomMean",
input_types=[Numeric],
return_type=Numeric)
from featuretools.primitives.base import make_agg_primitive
from featuretools.variable_types import Numeric
CustomMax = make_agg_primitive(lambda x: max(x),
name="CustomMax",
input_types=[Numeric],
return_type=Numeric)
from featuretools.primitives.base import make_agg_primitive
from featuretools.variable_types import Numeric
CustomSum = make_agg_primitive(lambda x: sum(x),
name="CustomSum",
input_types=[Numeric],
return_type=Numeric)