def test_no_crs(self):
df = gpd.GeoDataFrame({
'dem': [0, 1],
'rep': [1, 0],
'geometry': [
Polygon([(0, 0), (0, 0), (0, 0)]),
Polygon([(0, 0), (1, 1), (0, 1)])
]
})
with self.assertRaises(ValueError):
random_points_in_polygon(df, p=1)
def test_zero_area_polygon_error(self):
df = gpd.GeoDataFrame(
{
'dem': [0, 1],
'rep': [1, 0],
'geometry': [
Polygon([(0, 0), (0, 0), (0, 0)]),
Polygon([(0, 0), (1, 1), (0, 1)])
]
},
crs='esri:102010')
with self.assertRaises(ValueError):
random_points_in_polygon(df, p=1)
def test_random_points_in_polygon_seed(self):
df = gpd.GeoDataFrame(
{
'dem': [0, 1],
'rep': [1, 0],
'geometry': [
Polygon([(0, 0), (1, 1), (0, 1)]),
Polygon([(0, 0), (1, 1), (0, 1)])
]
},
crs='esri:102010')
result1 = random_points_in_polygon(df, p=0.5, random_seed=47)
result2 = random_points_in_polygon(df, p=0.5, random_seed=47)
pd.testing.assert_frame_equal(result1, result2)
def test_probability_value_large(self):
p = 0.29 # set the probability
delta = 0.015 # 1 standard error of mean
f = 8 # factor to be multiplied with delta
df = gpd.GeoDataFrame(
{
'dem': [10000],
'rep': [20000],
'geometry': [
Polygon([(0, 1), (1, 0), (0, 0), (1, 1)]),
]
},
crs='esri:102010')
district_test = gpd.GeoDataFrame(
{
'district': [1, 2, 3],
'geometry': [
Polygon([(-1, 0), (0, 1), (1, 0)]),
Polygon([(0, 1), (1, 1), (1, -1), (0, -1)]),
Polygon([(-1, 0), (1, 0), (-1, -1)])
]
},
crs='esri:102010')
df_random_points = random_points_in_polygon(df, p=0.29)
for party in ['dem', 'rep']:
relevant_dem_value = 1 if party == "dem" else 0
proportion = (df_random_points['dem']
== relevant_dem_value).sum() / df.loc[0, party]
assert abs(proportion - p) < f * delta
def test_probability_value_small(self):
p = 0.73 # set the probability
delta = 0.06 # a small number used to calculate interval for checking if generated dem proportion falls inside the range
f = 8 # factor to be multiplied with delta
df = gpd.GeoDataFrame(
{
'dem': [100],
'rep': [200],
'geometry': [
Polygon([(0, 1), (1, 0), (0, 0), (1, 1)]),
]
},
crs='esri:102010')
district_test = gpd.GeoDataFrame(
{
'district': [1, 2, 3],
'geometry': [
Polygon([(-1, 0), (0, 1), (1, 0)]),
Polygon([(0, 1), (1, 1), (1, -1), (0, -1)]),
Polygon([(-1, 0), (1, 0), (-1, -1)])
]
},
crs='esri:102010')
df_random_points = random_points_in_polygon(df, p=0.73)
for party in ['dem', 'rep']:
relevant_dem_value = 1 if party == "dem" else 0
proportion = (df_random_points['dem']
== relevant_dem_value).sum() / df.loc[0, party]
assert abs(proportion - p) < f * delta
def test_calculate_dislocation_simple_test(self):
df = gpd.GeoDataFrame(
{
'dem': [1, 1, 0, 0, 0, 0],
'rep': [0, 0, 1, 1, 1, 1],
'geometry': [
Polygon([(0, 1), (1, 0), (0, 0)]),
Polygon([(0, -1), (-1, 0), (0, 0)]),
Polygon([(0, 1), (-1, 0), (0, 0)]),
Polygon([(0, -1), (1, 0), (0, 0)]),
Polygon([(1, 0), (1, 1), (0, 1)]),
Polygon([(1, 0), (-1, -1), (0, -1)])
]
},
crs='esri:102010')
district_test = gpd.GeoDataFrame(
{
'district': [1, 2, 3],
'geometry': [
Polygon([(-1, 0), (0, 1), (1, 0)]),
Polygon([(0, 1), (1, 1), (1, -1), (0, -1)]),
Polygon([(-1, 0), (1, 0), (-1, -1)])
]
},
crs='esri:102010')
df_random_points = random_points_in_polygon(df, p=1)
df_voter_knn = calculate_voter_knn(df_random_points, k=3)
df_dislocation = calculate_dislocation(df_voter_knn, district_test)
def test_random_points_in_polygon_float_coordinates(self):
df = gpd.GeoDataFrame(
{
'dem': [0, 1],
'rep': [1, 0],
'geometry': [
Polygon([(-1.3, -1.0), (2.8, -3.1), (4.4, 7.9)]),
Polygon([(8.6, 10.5), (-5.3, -3.4), (6.2, 9.1)])
]
},
crs='esri:102010')
result = random_points_in_polygon(df, p=1)
benchmark = pd.Series([0, 1], name='dem')
pd.testing.assert_series_equal(result["dem"], benchmark)
def test_random_points_in_polygon_negative_coordinates(self):
df = gpd.GeoDataFrame(
{
'dem': [0, 1],
'rep': [1, 0],
'geometry': [
Polygon([(-1, -1), (2, -3), (4, 7)]),
Polygon([(8, 10), (-5, -3), (6, 9)])
]
},
crs='esri:102010')
result = random_points_in_polygon(df, p=1)
benchmark = pd.Series([0, 1], name='dem')
pd.testing.assert_series_equal(result["dem"], benchmark)
def test_random_points_in_polygon_simple_test(self):
df = gpd.GeoDataFrame(
{
'dem': [0, 1],
'rep': [1, 0],
'geometry': [
Polygon([(0, 0), (1, 1), (0, 1)]),
Polygon([(0, 0), (1, 1), (0, 1)])
]
},
crs='esri:102010')
result = random_points_in_polygon(df, p=1)
benchmark = pd.Series([0, 1], name='dem')
pd.testing.assert_series_equal(result["dem"], benchmark)