def __init__(self, vertices, color):
self.vertices = vertices
self.color = color
self.edges = [
Edge(*e, figure=self)
for e in zip(vertices,
list(vertices[1:]) + [vertices[0]])
]
self.inner_edges = [
Edge(*e, figure=self) for e in combinations(vertices, 2)
]
self.mpl_poly = MplPolygon(vertices, True, color=[1, 0, 0])
def plot(self, ax: GeoAxesSubplot, **kwargs) -> Artist:
"""Plotting function. All arguments are passed to the geometry"""
if "facecolor" not in kwargs:
kwargs["facecolor"] = "None"
if "edgecolor" not in kwargs:
kwargs["edgecolor"] = ax._get_lines.get_next_color()
if "projection" in ax.__dict__:
return ax.add_geometries([self.shape], crs=PlateCarree(), **kwargs)
else:
return ax.add_patch(
MplPolygon(list(self.shape.exterior.coords), **kwargs))
def plot(self):
# get coordinates in the Earth projection
xvec, yvec = self.projected(self._earthmap)
# create patch with coordinates
self._patch = MplPolygon(xy=np.array([xvec, yvec]).T,
closed=True,
linestyle=self._linestyle,
linewidth=self._linewidth,
fill=False,
color=self._color)
# add patch to plot
self._parent.get_axes().add_patch(self._patch)
# refresh plot
self._parent.draw()
def plot(self, ax: "GeoAxesSubplot", **kwargs: Any) -> "Artist":
"""Plotting function. All arguments are passed to the geometry"""
from cartopy.crs import PlateCarree
from matplotlib.patches import Polygon as MplPolygon
if "facecolor" not in kwargs:
kwargs["facecolor"] = "None"
if "edgecolor" not in kwargs:
kwargs["edgecolor"] = ax._get_lines.get_next_color()
if "projection" in ax.__dict__:
return ax.add_geometries([self.shape], crs=PlateCarree(), **kwargs)
else:
return ax.add_patch(
MplPolygon(list(self.shape.exterior.coords), **kwargs))
def plot(self, ax: GeoAxesSubplot, **kwargs) -> None: # coverage: ignore
flat = self.flatten()
if isinstance(flat, base.BaseMultipartGeometry):
for poly in flat:
# quick and dirty
sub = Airspace("", [ExtrudedPolygon(poly, 0, 0)])
sub.plot(ax, **kwargs)
return
if "facecolor" not in kwargs:
kwargs["facecolor"] = "None"
if "edgecolor" not in kwargs:
kwargs["edgecolor"] = ax._get_lines.get_next_color()
if "projection" in ax.__dict__:
ax.add_geometries([flat], crs=PlateCarree(), **kwargs)
else:
ax.add_patch(MplPolygon(list(flat.exterior.coords), **kwargs))
def union_polygons(list_of_polygons):
#accept few polygons and return their union, visualize them to make sure it's correct
fig = plt.figure(figsize=(10, 8))
ax = fig.add_subplot(111, axisbg='w', frame_on=False)
ax.xaxis.set_visible(False)
ax.yaxis.set_visible(False)
for spine in ax.spines.itervalues():
spine.set_visible(False)
m = Basemap(llcrnrlat=lllat,
urcrnrlat=urlat,
llcrnrlon=lllon,
urcrnrlon=urlon,
resolution='i',
projection='cyl')
m.drawmapboundary(fill_color='white')
#m.drawcoastlines(linewidth=0.2)
m.drawcountries(linewidth=0.2)
shp_info = m.readshapefile('../localism/data/us_states_st99/st99_d00',
'states',
drawbounds=False,
zorder=0)
regions = []
for shapedict, state in zip(m.states_info, m.states):
if shapedict['NAME'] in set(['California', 'Washington']):
regions.append(state)
polies = []
for r in regions:
poly = Polygon(r)
polies.append(poly)
poly = MultiPolygon(polies)
poly = unary_union(poly).convex_hull
lons, lats = poly.exterior.coords.xy
coords = np.array(zip(lons, lats))
poly = MplPolygon(coords, facecolor='gray', edgecolor='gray')
ax.add_patch(poly)
plt.show()
result = None
return result
def contour(coordinates,
scores,
world=False,
filename="contour",
do_contour=False,
**kwargs):
#with open('./data/coordinate_socres.pkl', 'wb') as fout:
# pickle.dump((coordinates, scores), fout)
with open('./data/coor_score_239.pkl', 'rb') as fin:
coordinates, scores = pickle.load(fin)
from matplotlib import rc
rc('font', **{'family': 'sans-serif', 'sans-serif': ['Helvetica']})
## for Palatino and other serif fonts use:
#rc('font',**{'family':'serif','serif':['Palatino']})
rc('text', usetex=True)
scores = np.array(scores)
lllat = 24.396308
lllon = -124.848974
urlat = 49.384358
urlon = -66.885444
if world:
lllat = -90
lllon = -180
urlat = 90
urlon = 180
fig = plt.figure(figsize=(2.5, 2))
grid_transform = kwargs.get('grid', False)
ax = fig.add_subplot(111, axisbg='w', frame_on=False)
grid_interpolation_method = 'nearest'
#scores = np.log(scores)
m = Basemap(llcrnrlat=lllat,
urcrnrlat=urlat,
llcrnrlon=lllon,
urcrnrlon=urlon,
resolution='i',
projection='cyl')
m.drawmapboundary(fill_color='white')
#m.drawcoastlines(linewidth=0.2)
m.drawcountries(linewidth=0.2)
if world:
m.drawstates(linewidth=0.2, color='lightgray')
#m.fillcontinents(color='white', lake_color='#0000ff', zorder=2)
#m.drawrivers(color='#0000ff')
#m.drawlsmask(land_color='gray',ocean_color="#b0c4de", lakes=True)
#m.drawcounties()
shp_info = m.readshapefile('./data/us_states_st99/st99_d00',
'states',
drawbounds=True,
zorder=0)
printed_names = []
ax = plt.gca()
ax.xaxis.set_visible(False)
ax.yaxis.set_visible(False)
for spine in ax.spines.itervalues():
spine.set_visible(False)
state_names_set = set(short_state_names.values())
mi_index = 0
wi_index = 0
for shapedict, state in zip(m.states_info, m.states):
if world: break
draw_state_name = True
if shapedict['NAME'] not in state_names_set: continue
short_name = short_state_names.keys()[short_state_names.values().index(
shapedict['NAME'])]
if short_name in printed_names and short_name not in ['MI', 'WI']:
continue
if short_name == 'MI':
if mi_index != 3:
draw_state_name = False
mi_index += 1
if short_name == 'WI':
if wi_index != 2:
draw_state_name = False
wi_index += 1
# center of polygon
x, y = np.array(state).mean(axis=0)
hull = ConvexHull(state)
hull_points = np.array(state)[hull.vertices]
x, y = hull_points.mean(axis=0)
if short_name == 'MD':
y = y - 0.5
x = x + 0.5
elif short_name == 'DC':
y = y + 0.1
elif short_name == 'MI':
x = x - 1
elif short_name == 'RI':
x = x + 1
y = y - 1
#poly = MplPolygon(state,facecolor='lightgray',edgecolor='black')
#x, y = np.median(np.array(state), axis=0)
# You have to align x,y manually to avoid overlapping for little states
if draw_state_name:
plt.text(x + .1, y, short_name, ha="center", fontsize=4)
#ax.add_patch(poly)
#pdb.set_trace()
printed_names += [
short_name,
]
mlon, mlat = m(*(coordinates[:, 1], coordinates[:, 0]))
# grid data
if do_contour:
numcols, numrows = 2000, 2000
xi = np.linspace(mlon.min(), mlon.max(), numcols)
yi = np.linspace(mlat.min(), mlat.max(), numrows)
xi, yi = np.meshgrid(xi, yi)
# interpolate
x, y, z = mlon, mlat, scores
#pdb.set_trace()
#zi = griddata(x, y, z, xi, yi)
zi = gd((mlon, mlat),
scores, (xi, yi),
method=grid_interpolation_method,
rescale=False)
#Remove the lakes and oceans
data = maskoceans(xi, yi, zi)
con = m.contourf(xi, yi, data, cmap=plt.get_cmap('YlOrRd'))
else:
cmap = plt.get_cmap('YlOrRd')
con = m.scatter(mlon, mlat, c=scores, s=3, cmap=cmap)
#con = m.contour(xi, yi, data, 3, cmap=plt.get_cmap('YlOrRd'), linewidths=1)
#con = m.contour(x, y, z, 3, cmap=plt.get_cmap('YlOrRd'), tri=True, linewidths=1)
#conf = m.contourf(x, y, z, 3, cmap=plt.get_cmap('coolwarm'), tri=True)
cbar = m.colorbar(con, location='right', pad="3%")
#plt.setp(cbar.ax.get_yticklabels(), visible=False)
#cbar.ax.tick_params(axis=u'both', which=u'both',length=0)
#cbar.ax.set_yticklabels(['low', 'high'])
#tick_locator = ticker.MaxNLocator(nbins=9)
#cbar.locator = tick_locator
#cbar.update_ticks()
cbar.ax.tick_params(labelsize=6)
cbar.ax.xaxis.set_tick_params(pad=0)
cbar.ax.yaxis.set_tick_params(pad=0)
cbar.set_label('error in km', size=8, labelpad=1)
for line in cbar.lines:
line.set_linewidth(20)
#read countries for world dataset with more than 100 number of users
with open('./data/country_count.json', 'r') as fin:
top_countries = set(json.load(fin))
world_shp_info = m.readshapefile(
'./data/CNTR_2014_10M_SH/Data/CNTR_RG_10M_2014',
'world',
drawbounds=False,
zorder=100)
for shapedict, state in zip(m.world_info, m.world):
if not world:
if shapedict['CNTR_ID'] not in ['CA', 'MX']: continue
else:
if shapedict['CNTR_ID'] in top_countries: continue
poly = MplPolygon(state, facecolor='gray', edgecolor='gray')
ax.add_patch(poly)
#plt.title('term: ' + word )
plt.tight_layout()
plt.savefig('./maps/' + filename + '.pdf', bbox_inches='tight')
plt.close()
del m
def map_words(coords, preds, vocab, map_dir, dataset_name):
"""
given the coords distributed over the map and
the unigram distribution over vocabulary pred,
contourf the logprob of a word over the map
with interpolation.
"""
lllat = 24.396308
lllon = -124.848974
urlat = 49.384358
urlon = -66.885444
if dataset_name == 'world-final':
lllat = -90
lllon = -180
urlat = 90
urlon = 180
grid_interpolation_method = 'cubic'
logging.info('interpolation: ' + grid_interpolation_method)
region_words = {
"the north":['braht','breezeway','bubbler','clout','davenport','euchre','fridge','hotdish','paczki','pop','sack','soda','toboggan','Yooper'],
"northeast":['brook','cellar','sneaker','soda'],
"New England":['grinder','packie','rotary','wicked'],
"Eastern New England":['bulkhead','Cabinet','frappe','hosey','intervale','jimmies','johnnycake','quahog','tonic'],
"Northern New England":['ayuh','creemee','dooryard','logan','muckle'],
"The Mid-Atlantic":['breezeway','hoagie','jawn','jimmies','parlor','pavement','shoobie','youze'],
"New York City Area":['bodega','dungarees','potsy','punchball','scallion','stoop','wedge'],
"The Midland":['hoosier'],
"The South":['banquette','billfold','chuck','commode','lagniappe','yankee','yonder'],
"The West":['davenport','Hella','snowmachine' ]
}
word_dialect = {}
with open('./data/geodare.cleansed.filtered.json', 'r') as fin:
for line in fin:
line = line.strip()
dialect_word = json.loads(line)
word_dialect[dialect_word['word']] = dialect_word['dialect']
#if os.path.exists(map_dir):
# shutil.rmtree(map_dir)
try:
os.mkdir(map_dir)
except:
logging.info('map_dir %s exists or can not be created.')
#pick some words to map including some known dialect words
#some DARE words and some words that are not evenly distributed
topk_words = []
for words in region_words.values():
topk_words.extend(words)
topk_words.extend(word_dialect.keys())
dialect_words = ['hella', 'yall', 'jawn', 'paczki', 'euchre', 'brat', 'toboggan', 'brook', 'grinder', 'yinz', 'youze', 'yeen']
topk_words.extend(dialect_words)
custom_words = ['springfield', 'columbia', 'n***a', 'niqqa', 'bamma', 'cooter', 'britches', 'yapper', 'younguns', 'hotdish',
'schnookered', 'bubbler', 'betcha', 'dontcha']
topk_words.extend(custom_words)
vocabset = set(vocab)
dare_in_vocab = set(word_dialect.keys()) & vocabset
logging.info('%d DARE words, %d in vocab' %(len(word_dialect), len(dare_in_vocab)))
add_local_words = True
if add_local_words:
ne_file = './dumps/ne_' + dataset_name + '.json'
with codecs.open(ne_file, 'r', encoding='utf-8') as fout:
NEs = json.load(fout)
NEs = NEs['nes']
local_words = get_local_words(preds, vocab, NEs=NEs, k=500)
logging.info(local_words)
topk_words.extend(local_words[0:20])
add_cities = False
if add_cities:
with open('./data/cities.json', 'r') as fin:
cities = json.load(fin)
cities = cities[0:100]
for city in cities:
name = city['city'].lower()
topk_words.append(name)
wi = 0
for word in topk_words:
if word in vocabset:
fig = plt.figure(figsize=(5, 4))
ax = fig.add_subplot(111, axisbg='w', frame_on=False)
logging.info('%d mapping %s' %(wi, word))
wi += 1
index = vocab.index(word)
scores = np.log(preds[:, index])
m = Basemap(llcrnrlat=lllat,
urcrnrlat=urlat,
llcrnrlon=lllon,
urcrnrlon=urlon,
resolution='i', projection='cyl')
'''
m = Basemap(llcrnrlon=-119,llcrnrlat=22,urcrnrlon=-64,urcrnrlat=49,
projection='lcc',lat_1=33,lat_2=45,lon_0=-95, resolution='i')
'''
m.drawmapboundary(fill_color = 'white')
#m.drawcoastlines(linewidth=0.2)
m.drawcountries(linewidth=0.2)
if dataset_name != 'world-fianl':
m.drawstates(linewidth=0.2, color='lightgray')
#m.fillcontinents(color='white', lake_color='#0000ff', zorder=2)
#m.drawrivers(color='#0000ff')
#m.drawlsmask(land_color='gray',ocean_color="#b0c4de", lakes=True)
#m.drawcounties()
shp_info = m.readshapefile('./data/us_states_st99/st99_d00','states',drawbounds=True, zorder=0)
printed_names = []
ax = plt.gca()
ax.xaxis.set_visible(False)
ax.yaxis.set_visible(False)
for spine in ax.spines.itervalues():
spine.set_visible(False)
state_names_set = set(short_state_names.values())
mi_index = 0
wi_index = 0
for shapedict,state in zip(m.states_info, m.states):
if dataset_name == 'world-final': break
draw_state_name = True
if shapedict['NAME'] not in state_names_set: continue
short_name = short_state_names.keys()[short_state_names.values().index(shapedict['NAME'])]
if short_name in printed_names and short_name not in ['MI', 'WI']:
continue
if short_name == 'MI':
if mi_index != 3:
draw_state_name = False
mi_index += 1
if short_name == 'WI':
if wi_index != 2:
draw_state_name = False
wi_index += 1
# center of polygon
x, y = np.array(state).mean(axis=0)
hull = ConvexHull(state)
hull_points = np.array(state)[hull.vertices]
x, y = hull_points.mean(axis=0)
if short_name == 'MD':
y = y - 0.5
x = x + 0.5
elif short_name == 'DC':
y = y + 0.1
elif short_name == 'MI':
x = x - 1
elif short_name == 'RI':
x = x + 1
y = y - 1
#poly = MplPolygon(state,facecolor='lightgray',edgecolor='black')
#x, y = np.median(np.array(state), axis=0)
# You have to align x,y manually to avoid overlapping for little states
if draw_state_name:
plt.text(x+.1, y, short_name, ha="center", fontsize=8)
#ax.add_patch(poly)
#pdb.set_trace()
printed_names += [short_name,]
mlon, mlat = m(*(coords[:,1], coords[:,0]))
# grid data
numcols, numrows = 1000, 1000
xi = np.linspace(mlon.min(), mlon.max(), numcols)
yi = np.linspace(mlat.min(), mlat.max(), numrows)
xi, yi = np.meshgrid(xi, yi)
# interpolate
x, y, z = mlon, mlat, scores
#pdb.set_trace()
#zi = griddata(x, y, z, xi, yi)
zi = gd(
(mlon, mlat),
scores,
(xi, yi),
method=grid_interpolation_method, rescale=False)
#Remove the lakes and oceans
data = maskoceans(xi, yi, zi)
con = m.contourf(xi, yi, data, cmap=plt.get_cmap('YlOrRd'))
#con = m.contour(xi, yi, data, 3, cmap=plt.get_cmap('YlOrRd'), linewidths=1)
#con = m.contour(x, y, z, 3, cmap=plt.get_cmap('YlOrRd'), tri=True, linewidths=1)
#conf = m.contourf(x, y, z, 3, cmap=plt.get_cmap('coolwarm'), tri=True)
cbar = m.colorbar(con,location='right',pad="2%")
#plt.setp(cbar.ax.get_yticklabels(), visible=False)
#cbar.ax.tick_params(axis=u'both', which=u'both',length=0)
#cbar.ax.set_yticklabels(['low', 'high'])
tick_locator = ticker.MaxNLocator(nbins=9)
cbar.locator = tick_locator
cbar.update_ticks()
cbar.ax.tick_params(labelsize=11)
cbar.ax.yaxis.set_tick_params(pad=2)
cbar.set_label('logprob', size=11)
for line in cbar.lines:
line.set_linewidth(10)
#read countries for world dataset with more than 100 number of users
with open('./data/country_count.json', 'r') as fin:
top_countries = set(json.load(fin))
world_shp_info = m.readshapefile('./data/CNTR_2014_10M_SH/Data/CNTR_RG_10M_2014','world',drawbounds=False, zorder=100)
for shapedict,state in zip(m.world_info, m.world):
if dataset_name != 'world-final':
if shapedict['CNTR_ID'] not in ['CA', 'MX']: continue
else:
if shapedict['CNTR_ID'] in top_countries: continue
poly = MplPolygon(state,facecolor='gray',edgecolor='gray')
ax.add_patch(poly)
#plt.title('term: ' + word )
plt.tight_layout()
filename = '{}{}_{}.pdf'.format(map_dir, word.encode('utf-8'), grid_interpolation_method)
plt.savefig(filename, bbox_inches='tight')
plt.close()
del m