def test_normalize():
index = pandas.MultiIndex.from_tuples([(0, 1), (0, 2), (1, 2), (1, 3),
(1, 4), (2, 4)])
weights = pandas.Series([10, 20, 25, 15, 0, 30], index=index)
expected = pandas.Series([1 / 3, 2 / 3, 25 / 40, 15 / 40, 0, 1],
index=index)
assert (maup.normalize(weights) == expected).all(0)
def test_crop_to():
blocks = geopandas.read_file("zip://./examples/blocks.zip")
old_precincts = geopandas.read_file("zip://./examples/precincts.zip")
new_precincts = geopandas.read_file("zip://./examples/new_precincts.zip")
columns = ["SEN18D", "SEN18R"]
# Calculate without cropping
pieces = maup.intersections(old_precincts, new_precincts, area_cutoff=0)
weights = blocks["TOTPOP"].groupby(maup.assign(blocks, pieces)).sum()
weights = maup.normalize(weights, level=0)
new_precincts[columns] = maup.prorate(pieces,
old_precincts[columns],
weights=weights)
# Calculate with cropping
old_precincts["geometries"] = maup.crop_to(old_precincts, new_precincts)
new_precincts_cropped = new_precincts.copy()
pieces = maup.intersections(old_precincts,
new_precincts_cropped,
area_cutoff=0)
weights = blocks["TOTPOP"].groupby(maup.assign(blocks, pieces)).sum()
weights = maup.normalize(weights, level=0)
new_precincts_cropped[columns] = maup.prorate(pieces,
old_precincts[columns],
weights=weights)
assert new_precincts_cropped.area.sum() != new_precincts.area.sum()
diff_sum = 0
for col in columns:
diff = new_precincts_cropped[col].sum() - new_precincts[col].sum()
assert diff >= 0
diff_sum += diff
# Ideally this would be strictly positive (which would mean less votes are lost after cropping)
# but crop_to doesn't resolve the missing votes errors yet.
assert diff_sum >= 0
def test_example_case():
blocks = geopandas.read_file("zip://./examples/blocks.zip")
old_precincts = geopandas.read_file("zip://./examples/precincts.zip")
new_precincts = geopandas.read_file("zip://./examples/new_precincts.zip")
columns = ["SEN18D", "SEN18R"]
# Include area_cutoff=0 to ignore any intersections with no area,
# like boundary intersections, which we do not want to include in
# our proration.
pieces = intersections(old_precincts, new_precincts, area_cutoff=0)
# Weight by prorated population from blocks
weights = blocks["TOTPOP"].groupby(assign(blocks, pieces)).sum()
weights = normalize(weights, level=0)
# Use blocks to estimate population of each piece
new_precincts[columns] = prorate(pieces, old_precincts[columns], weights=weights)
assert (new_precincts[columns] > 0).sum().sum() > len(new_precincts) / 2
for col in columns:
assert abs(new_precincts[col].sum() - old_precincts[col].sum()) / old_precincts[col].sum() < 0.1
il1990 = gpd.read_file(chicago_1990_file)
blocks = gpd.read_file(chicago_blocks_file)
il2000.crs = il2010.crs
il1990.crs = il2010.crs
blocks.to_crs(il2010.crs, inplace=True)
for c in columns:
il2010[c] = il2010[c].astype(int)
il2000[c] = il2000[c].astype(int)
il1990[c] = il1990[c].astype(int)
pieces2000 = maup.intersections(il2000, il2010, area_cutoff=0)
pieces1990 = maup.intersections(il1990, il2010, area_cutoff=0)
weights2000 = blocks["TOTPOP"].groupby(maup.assign(blocks, pieces2000)).sum()
weights1990 = blocks["TOTPOP"].groupby(maup.assign(blocks, pieces1990)).sum()
weights2000 = maup.normalize(weights2000, level=0)
weights1990 = maup.normalize(weights1990, level=0)
il2010[columns2000] = maup.prorate(pieces2000, il2000[columns], weights=weights2000)
il2010[columns1990] = maup.prorate(pieces1990, il1990[columns], weights=weights1990)
il2010.plot(column=il2010["TOTPOP_2000"].isna())
plt.show()
print(il2010["NH_BLACK_2000"])
print(il2010["TOTPOP_2000"])
def relentropy(df, races, totpop_col):
totpop = sum(x for x in df[totpop_col] if not isnan(x))
res = 0
for j in range(0, df.shape[0]):