Skip to content

Fix issues with auto_entityset #21

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 10 commits into
base: main
Choose a base branch
from
Open

Fix issues with auto_entityset #21

Show file tree
Hide file tree
Changes from 6 commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
63aa776
fix issue with nan values in approximate_dependencies
Mar 27, 2020
f6ee78e
add test for catching improper dropping of cols needed for relationships
Mar 30, 2020
5efa2d5
Merge branch 'master' of https://github.com/FeatureLabs/autonormalize…
Apr 2, 2020
4fa5693
Merge branch 'master' of
May 19, 2020
23b93fb
Merge branch 'main' into issue19
gsheni Jan 10, 2022
ceeadc0
Merge branch 'main' into issue19
rwedge May 9, 2022
04f7f71
update release notes
May 11, 2022
548ef6c
remove comment
May 11, 2022
0842384
remove temporary test
May 11, 2022
036c1a0
lint fix
May 11, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 8 additions & 4 deletions autonormalize/dfd.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,7 +1,7 @@
from functools import partial
from itertools import combinations

import numpy
import pandas as pd
from tqdm import tqdm

from .classes import DfdDependencies, LHSs, Masks, Node
Expand Down Expand Up @@ -359,7 +359,6 @@ def approximate_dependencies(lhs_set, rhs, df, accuracy, masks):
acc = 0

for index, row in indicator.iterrows():

mask = None
for attr in lhs_set:

Expand All @@ -368,14 +367,19 @@ def approximate_dependencies(lhs_set, rhs, df, accuracy, masks):
if df[attr].dtypes.name == 'datetime64[ns]':
m = df[attr] == row[attr]
else:
m = df[attr].values == row[attr]
if pd.isna(row[attr]):
m = df[attr].isnull()
else:
m = df[attr].values == row[attr]
masks.add_mask(attr, row[attr], m)
if mask is None:
mask = m
else:
mask = mask & m
options = df[mask]
_, unique_counts = numpy.unique(options[rhs].to_numpy(), return_counts=True)

# _, unique_counts = np.unique(options[rhs].to_numpy(), return_counts=True)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

should this line be removed?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yep, removed this and another temporary test that I added trying to replicate the problem that was present in #19.

unique_counts = options[rhs].value_counts()
acc += unique_counts.sum() - unique_counts.max()
if acc > limit:
return False
Expand Down
6 changes: 4 additions & 2 deletions autonormalize/normalize.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -76,6 +76,7 @@ def make_indexes(depdf):
Arguments:
depdf (DepDF) : depDF to make indexes for
"""

prim_key = depdf.deps.get_prim_key()

if len(prim_key) > 1:
Expand Down Expand Up @@ -103,8 +104,9 @@ def make_indexes(depdf):

for index in indices[name]:
add[index] = new_val

depdf.parent.df.drop(columns=prim_key, inplace=True)
# Don't drop a column if it is needed in another parent relationship
to_drop = [key for key in prim_key if key not in depdf.parent.deps.serialize().keys()]
depdf.parent.df.drop(columns=to_drop, inplace=True)
depdf.parent.df.insert(len(depdf.parent.df.columns), '_'.join(prim_key), add)

for child in depdf.children:
Expand Down
44 changes: 26 additions & 18 deletions autonormalize/tests/test_dfd.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,6 @@
import os

import numpy as np
import pandas as pd

from autonormalize import dfd
Expand Down Expand Up @@ -73,21 +74,28 @@ def test_compute_partitions():
assert not dfd.compute_partitions(df, 'c', frozenset(['a', 'b']), {}, 0.96, mask)


# def test_approximate_dependencies():
# mask = dfd.Masks(['a', 'b', 'c'])
# a = [6, 2, 3, 7, 8, 1, 0, 2, 0, 3, 6, 0, 4, 6, 8, 7, 6, 8, 1, 5, 1, 3, 3, 0, 0, 4, 5, 5, 7, 0, 8, 2, 4, 7, 0, 0, 6, 4, 6, 8]
# # b = [int(x%2 == 0) for x in a]
# b = [1, 1, 0, 0, 1, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1]
# # c = [(a[i] + b[i])<4 for i in range(40)]
# c = [False, True, True, False, False, True, True, True, True, True, False, True, False, False, False, False, False, False, True, False, True, True, True, True, True, False, False, False, False, True, False, True, False, False, True, True, False, False, False, False]
# df = pd.DataFrame({'a': a, 'b': b, 'c': c})
# assert dfd.approximate_dependencies([0, 1], 2, df, 1.00, mask, 0.90)
# assert dfd.approximate_dependencies(set([0, 1]), 2, df, .90, mask, 0.90)
# c[0] = True
# df = pd.DataFrame({'a': a, 'b': b, 'c': c})
# assert dfd.approximate_dependencies([0, 1], 2, df, .97, mask, 0.90)
# assert not dfd.approximate_dependencies(set([0, 1]), 2, df, .98, mask, 0.90)
# c[35] = False
# df = pd.DataFrame({'a': a, 'b': b, 'c': c})
# assert dfd.approximate_dependencies([0, 1], 2, df, .95, mask, 0.90)
# assert not dfd.approximate_dependencies([0, 1], 2, df, .96, mask, 0.90)
def test_approximate_dependencies():
mask = dfd.Masks(['a', 'b', 'c'])
a = [6, 2, 3, 7, 8, 1, 0, 2, 0, 3, 6, 0, 4, 6, 8, 7, 6, 8, 1, 5, 1, 3, 3, 0, 0, 4, 5, 5, 7, 0, 8, 2, 4, 7, 0, 0, 6, 4, 6, 8]
b = [1, 1, 0, 0, 1, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1]
c = [False, True, True, False, False, True, True, True, True, True, False, True, False, False, False, False, False, False, True, False, True, True, True, True, True, False, False, False, False, True, False, True, False, False, True, True, False, False, False, False]
df = pd.DataFrame({'a': a, 'b': b, 'c': c})
assert dfd.approximate_dependencies(['a', 'b'], 'c', df, 1.00, mask)
assert dfd.approximate_dependencies(['a', 'b'], 'c', df, .90, mask)
c[0] = True
df = pd.DataFrame({'a': a, 'b': b, 'c': c})
assert dfd.approximate_dependencies(['a', 'b'], 'c', df, .97, mask)
assert not dfd.approximate_dependencies(['a', 'b'], 'c', df, .98, mask)
c[35] = False
df = pd.DataFrame({'a': a, 'b': b, 'c': c})
assert dfd.approximate_dependencies(['a', 'b'], 'c', df, .95, mask)
assert not dfd.approximate_dependencies(['a', 'b'], 'c', df, .96, mask)


def test_approximate_dependencies_with_nan():
mask = dfd.Masks(['a', 'b', 'c'])
a = [np.nan, 2, 3, 7, 8, 1, 0, 2, 0, 3, np.nan, 0, 4, np.nan, 8, 7, np.nan, 8, 1, 5, 1, 3, 3, 0, 0, 4, 5, 5, 7, 0, 8, 2, 4, 7, 0, 0, np.nan, 4, np.nan, 8]
b = [1, 1, 0, 0, 1, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1]
c = [True, True, True, False, False, True, True, True, True, True, False, True, False, False, False, False, False, False, True, False, True, True, True, True, True, False, False, False, False, True, False, True, False, False, True, True, False, False, False, False]
df = pd.DataFrame({'a': a, 'b': b, 'c': c})
assert dfd.approximate_dependencies(['a', 'b'], 'c', df, 0.9, mask)
48 changes: 48 additions & 0 deletions autonormalize/tests/test_normalize.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,8 @@
import numpy as np
import pandas as pd
from pandas.testing import assert_frame_equal

import autonormalize as an
from autonormalize import classes, normalize


Expand Down Expand Up @@ -178,3 +180,49 @@ def test_make_indexes():
# Make sure new column names are sorted
assert 'hemisphere_month' in new_dfs[0].columns
assert 'hemisphere_month' in new_dfs[1].columns


def test_make_indexes_improper_column_drop():
df_dict = {'Id': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
'MSSubClass': [90, 60, 90, 90, 20, 50, 80, 20, 60, 20, 20],
'MSZoning': ['RL', 'RL', 'RL', 'RL', 'RL', 'RM', 'RL', 'RL', 'RL', 'RL', 'RL'],
'LotFrontage': [55.0, np.nan, 42.0, 100.0, np.nan, 98.0, 70.0, 85.0, 65.0, 78.0, 60.0],
'LotArea': [12640, 8755, 7711, 25000, 14375, 8820, 8163, 14536, 14006, 9360, 7200],
'Alley': [np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan],
'LotShape': ['IR1', 'IR1', 'IR1', 'Reg', 'IR1', 'Reg', 'Reg', 'Reg', 'IR1', 'Reg', 'Reg'],
'LandContour': ['Lvl', 'Lvl', 'Lvl', 'Low', 'Lvl', 'Lvl', 'Lvl', 'Lvl', 'Lvl', 'Lvl', 'Lvl'],
'Utilities': ['AllPub', 'AllPub', 'AllPub', 'AllPub', 'NoSeWa', 'AllPub', 'AllPub', 'AllPub', 'AllPub', 'AllPub', 'AllPub']}
df = pd.DataFrame(df_dict)

deps = classes.Dependencies({'Id': [['LotArea']],
'MSSubClass': [['LotArea'], ['LotFrontage', 'Utilities'], ['Id']],
'MSZoning': [['LotFrontage'], ['LotArea'], ['MSSubClass'], ['Id']],
'LotFrontage': [['LotArea'], ['Id']], 'LotArea': [['Id']],
'Alley': [['LotFrontage'], ['LandContour'], ['Utilities'], ['MSSubClass'], ['Id'], ['MSZoning'], ['LotArea'], ['LotShape']],
'LotShape': [['LotFrontage'], ['MSSubClass', 'Utilities', 'LandContour'], ['LotArea'], ['Id']],
'LandContour': [['LotFrontage'], ['MSSubClass', 'LotShape'], ['LotArea'], ['Id']],
'Utilities': [['MSSubClass', 'LotShape'], ['LotArea'], ['MSSubClass', 'LotFrontage'], ['Id']]}, ['id'])

depdf = normalize.DepDF(deps, df, deps.get_prim_key())
normalize.normalize_dataframe(depdf)
normalize.make_indexes(depdf)
new_dfs = depdf.return_dfs()

assert 'MSSubClass' in new_dfs[0].columns


def test_issue19():
df_dict = {'Id': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
'MSSubClass': [90, 60, 90, 90, 20, 50, 80, 20, 60, 20, 20],
'MSZoning': ['RL', 'RL', 'RL', 'RL', 'RL', 'RM', 'RL', 'RL', 'RL', 'RL', 'RL'],
'LotFrontage': [55.0, np.nan, 42.0, 100.0, np.nan, 98.0, 70.0, 85.0, 65.0, 78.0, 60.0],
'LotArea': [12640, 8755, 7711, 25000, 14375, 8820, 8163, 14536, 14006, 9360, 7200],
'Alley': [np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan],
'LotShape': ['IR1', 'IR1', 'IR1', 'Reg', 'IR1', 'Reg', 'Reg', 'Reg', 'IR1', 'Reg', 'Reg'],
'LandContour': ['Lvl', 'Lvl', 'Lvl', 'Low', 'Lvl', 'Lvl', 'Lvl', 'Lvl', 'Lvl', 'Lvl', 'Lvl'],
'Utilities': ['AllPub', 'AllPub', 'AllPub', 'AllPub', 'NoSeWa', 'AllPub', 'AllPub', 'AllPub', 'AllPub', 'AllPub', 'AllPub']}

df = pd.DataFrame(df_dict)

es = an.auto_entityset(df, accuracy=1.0, name="es")
print(es)