def spatial_density_all_time_all_crimes():
poly = get_camden_region()
camden_mpoly = geodjango_to_shapely([get_camden_region()])
oo = osm.OsmRendererBase(poly, buffer=100)
cbg = CadByGrid()
a = cbg.all_time_aggregate()
fig = plt.figure(figsize=(8, 8))
ax = fig.add_subplot(111, projection=ccrs.OSGB())
ax.set_extent([523000, 533000, 179000, 190000], ccrs.OSGB())
ax.background_patch.set_visible(False)
oo.render(ax)
a = a.sum(axis=1)
cmap = mpl.cm.Reds
norm = mpl.colors.Normalize()
norm.autoscale(a)
cax = mpl.colorbar.make_axes(ax,
location='bottom',
pad=0.02,
fraction=0.05,
shrink=0.9)
cbar = mpl.colorbar.ColorbarBase(cax[0],
cmap=cmap,
norm=norm,
orientation='horizontal')
sm = mpl.cm.ScalarMappable(norm=norm, cmap=cmap)
for j in range(cbg.m):
val = a.values[j]
fc = sm.to_rgba(val) if val else 'none'
ax.add_geometries(geodjango_to_shapely([cbg.grid[j].mpoly]),
ccrs.OSGB(),
facecolor=fc,
alpha=0.3)
ax.add_geometries(camden_mpoly,
ccrs.OSGB(),
facecolor='none',
edgecolor='black')
plt.show()
def get_camden_wards(as_shapely=True):
camden = models.Division.objects.get(type='borough', name__iexact='camden')
wards = models.Division.objects.filter(type='ward',
mpoly__intersects=camden.mpoly)
polys = []
for t in wards:
if t.mpoly.intersection(camden.mpoly).area / t.mpoly.area > 1e-3:
polys.append(t.mpoly.simplify())
if as_shapely:
polys = [geodjango_to_shapely(p) for p in polys]
return polys
def plot_domains(name=None, ax=None, set_axis=True):
if name is None and ax is not None:
raise AttributeError("If no single domain is specified, we have to make new figures for each plot")
city = geodjango_to_shapely(
models.ChicagoDivision.objects.get(name='ChicagoFilled').mpoly.simplify(0)
)
domains = get_chicago_side_polys(as_shapely=True)
if name is not None:
plot_domain(city, domains[name], ax=ax, set_axis=set_axis)
else:
for k, v in domains.items():
plot_domain(city, v)
def __init__(self, grid=None, **kwargs):
# defer dedupe until after spatial aggregation
super(CadSpatialGrid, self).__init__(**kwargs)
# load grid or use one provided
self.grid = grid or models.Division.objects.filter(
type='cad_250m_grid')
self.shapely_grid = [
geodjango_to_shapely(x.mpoly)[0] for x in self.grid
]
# self.shapely_grid = pandas.Series([geodjango_to_shapely([x.mpoly]) for x in self.grid],
# index=[x.name for x in self.grid])
# gridded data
self.data = self.aggregate()
def __init__(self, data, t0, data_index=None, grid=None):
"""
:param data: N x 3 np.ndarray or DataArray (t, x, y), all float
:param data_index: length N array containing indices corresponding to data. If not supplied, lookup index used.
:param t0: datetime corresponding to first record
:param grid: optional list of grid (multi)polygons, default is CAD 250m grid
:return:
"""
self.grid = grid or [
t.mpoly
for t in models.Division.objects.filter(type='cad_250m_grid')
]
self.shapely_grid = [geodjango_to_shapely(x)[0] for x in self.grid]
# preliminary cad filter
self.data = data
self.t0 = t0
self.data_index = np.array(
data_index) if data_index is not None else np.arange(len(data))
self.dates, self.indices = self.compute_array()
def get_camden_region(as_shapely=False):
poly = models.Division.objects.get(type='borough', name__iexact='camden')
poly = poly.mpoly.simplify() # type Polygon
if as_shapely:
poly = geodjango_to_shapely(poly)
return poly
def local_morans_i():
nicl_numbers = [1, 3, (6, 7)]
short_names = ['Violence', 'Burglary dwelling', 'Theft of/from vehicle']
camden_mpoly = geodjango_to_shapely(
[models.Division.objects.get(name='Camden', type='borough').mpoly])
cbg = CadByGrid(nicl_numbers=nicl_numbers)
a = cbg.all_time_aggregate()
W = rook_boolean_connectivity(cbg.grid)
fig = plt.figure(figsize=(15, 6))
axes = [
fig.add_subplot(1, cbg.l, i + 1, projection=ccrs.OSGB())
for i in range(cbg.l)
]
fig.subplots_adjust(left=0.05,
right=0.95,
bottom=0.05,
top=0.95,
wspace=0.03,
hspace=0.01)
local_i = []
standard_local_i = []
max_li = 0.
min_li = 1e6
for i in range(cbg.l):
ds = a[cbg.nicl_names[i]]
this_res = lmi(ds, W)
local_i.append(this_res)
standard_local_i.append(this_res / this_res.std())
max_li = max(max_li, max(standard_local_i[i]))
min_li = min(min_li, min(standard_local_i[i]))
for i in range(cbg.l):
ax = axes[i]
ax.set_title(short_names[i])
ax.set_extent([523000, 533000, 179000, 190000], ccrs.OSGB())
ax.background_patch.set_visible(False)
# ax.outline_patch.set_visible(False)
cmap = mpl.cm.jet
norm = mpl.colors.Normalize(vmin=min_li, vmax=max_li)
# norm.autoscale(local_i[i])
cax = mpl.colorbar.make_axes(ax,
location='bottom',
pad=0.02,
fraction=0.05,
shrink=0.9)
cbar = mpl.colorbar.ColorbarBase(cax[0],
cmap=cmap,
norm=norm,
orientation='horizontal')
sm = mpl.cm.ScalarMappable(norm=norm, cmap=cmap)
for j in range(cbg.m):
val = standard_local_i[i].values[j]
fc = sm.to_rgba(val) if val else 'none'
ax.add_geometries(geodjango_to_shapely([cbg.grid[j].mpoly]),
ccrs.OSGB(),
facecolor=fc)
ax.add_geometries(camden_mpoly,
ccrs.OSGB(),
facecolor='none',
edgecolor='black')
plt.show()
return local_i
def spatial_density_weekday_evening():
nicl_numbers = [3, 6, 10]
short_names = ['Burglary Dwelling', 'Veh Theft', 'Crim Damage']
camden_mpoly = geodjango_to_shapely(
[models.Division.objects.get(name='Camden', type='borough').mpoly])
cbg = CadByGrid(nicl_numbers=nicl_numbers)
a = cbg.weekday_weekend_aggregate()
b = cbg.daytime_evening_aggregate()
# combine
for x in b.items:
a[x] = b[x]
b = a.transpose(2, 0, 1)
for i in range(cbg.l): # crime types
df = b.iloc[i]
n = df.shape[0]
fig = plt.figure(figsize=(20, 6))
axes = [
fig.add_subplot(1, n, t + 1, projection=ccrs.OSGB())
for t in range(n)
]
fig.subplots_adjust(left=0.05,
right=0.95,
bottom=0.05,
top=0.95,
wspace=0.03,
hspace=0.01)
for j in range(n):
ax = axes[j]
ds = df.iloc[j]
ax.set_title(ds.name)
ax.set_extent([523000, 533000, 179000, 190000], ccrs.OSGB())
ax.background_patch.set_visible(False)
# ax.outline_patch.set_visible(False)
cmap = mpl.cm.cool
norm = mpl.colors.Normalize()
norm.autoscale(ds)
cax = mpl.colorbar.make_axes(ax,
location='bottom',
pad=0.02,
fraction=0.05,
shrink=0.9)
cbar = mpl.colorbar.ColorbarBase(cax[0],
cmap=cmap,
norm=norm,
orientation='horizontal')
sm = mpl.cm.ScalarMappable(norm=norm, cmap=cmap)
for j in range(cbg.m):
val = ds.values[j]
fc = sm.to_rgba(val) if val else 'none'
ax.add_geometries(geodjango_to_shapely([cbg.grid[j].mpoly]),
ccrs.OSGB(),
facecolor=fc)
ax.add_geometries(camden_mpoly,
ccrs.OSGB(),
facecolor='none',
edgecolor='black')
plt.show()