def get_shapes(self, max_area, max_distance_meters):
"return the shapes [census tracts] that touch the center of max_area + max_distance"
query = {
'geometry': {
'$near':
SON([('$geometry',
SON([('type', 'Point'),
('coordinates', [longitude, latitude])])),
('$maxDistance', max_distance_meters)])
}
}
for shp in self.db.GENZ2010_140.find(query):
# add population to properties
p = shp['properties']
# for later use
p['label'] = "{} {} [{}]".format(
state_name(p['STATE']), county_name(p['STATE'], p['COUNTY']),
p['TRACT'])
pop = {}
# lookup population and area
raw_pop = int(
self.census.sf1.state_county_tract("P0010001", p['STATE'],
p['COUNTY'],
p['TRACT'])[0]["P0010001"])
pop['raw'] = raw_pop
#print("raw population: {}".format(raw_pop))
# compute area in our radius
county_shape = shape(shp['geometry'])
isect = county_shape.intersection(max_area)
# save the intersection for later use.
shp['intersection'] = isect
pop['area'] = county_shape.area
frac_contained = float(isect.area) / float(county_shape.area)
pop['frac_contained'] = frac_contained
#print("contained: {}".format(frac_contained))
# compute effective population
#print("contained population: {}".format(frac_contained * raw_pop))
pop['effective'] = frac_contained * raw_pop
p['population'] = pop
# filter out areas with very low population
# the mongodb geo query may be not exact
if frac_contained * raw_pop < 0.1:
print("WARN: clipping: {}".format(p['label']))
pprint(pop)
continue
yield shp
def plot_shapes(ax,
shapes,
filled=False,
show_states=False,
fc=None,
alpha=0.16):
# have to buffer the results to see how many colors to generate
if fc == None:
if filled:
color_list = plt.cm.Dark2(np.linspace(0, 1, len(shapes)))
fc = lambda x: color_list[x]
else:
fc = lambda x: 'none'
states = set()
geoid = {}
# get the shapes covered by the coverage area.
for i, shp in enumerate(shapes):
p = shp['properties']
#track states we hit
if p['STATE'] not in states:
states.add(p['STATE'])
# also store the goeid so we dont plot again later
geoid[p['GEO_ID']] = shp
if p['LSAD'] == 'Tract':
label = "{} {} [{}]".format(state_name(p['STATE']),
county_name(p['STATE'], p['COUNTY']),
p['TRACT'])
else:
label = p['NAME']
#print ("-"*10 + label + "-"*10)
#pprint(p['population'])
ec = 'black'
lw = 1
if 'population' in p:
if p['population']['effective'] < 0.01:
ec = 'blue'
lw = 2
#was fc = color_list[i],
patches = make_patch(shp['geometry'],
fc=fc(i),
lw=lw,
ec=ec,
label=label,
alpha=alpha)
for p in patches:
ax.add_patch(p)
def _add_pop_to_shape_intersct(self, shp, max_area):
"add the population inside the intersection of shp and max_area, shp is assumed to be a census tract."
# add population to properties
p = shp['properties']
#print(p)
# for later use
p['label'] = "{} {} [{}]".format(state_name(p['STATE']),
county_name(p['STATE'], p['COUNTY']),
p['TRACT'])
pop = {}
# lookup population and area try mongo first.
searchdoc = {
'state': p["STATE"],
'county': p['COUNTY'],
'tract': p['TRACT'],
'variable.P0010001': {
'$exists': True
}
}
#print("searchdoc: ",end="")
#pprint(searchdoc)
cdoc = self.census_col.find_one(searchdoc)
if cdoc != None:
#print("Hit.")
raw_pop = cdoc['variable']['P0010001']
else:
raw_pop = int(
self.census.sf1.state_county_tract("P0010001", p['STATE'],
p['COUNTY'],
p['TRACT'])[0]["P0010001"])
insdoc = {
'state': p["STATE"],
'county': p['COUNTY'],
'tract': p['TRACT'],
'variable': {
'P0010001': raw_pop
}
}
#print("inserting: ", end="")
#pprint(insdoc)
self.census_col.insert_one(insdoc)
pop['raw'] = raw_pop
#print("raw population: {}".format(raw_pop))
county_shape = shape(shp['geometry'])
if max_area != None:
# compute area in our radius
isect = county_shape.intersection(max_area)
union = county_shape.union(max_area)
# save the intersection for later use.
shp['intersection'] = mapping(isect)
shp['union'] = mapping(union)
pop['area'] = county_shape.area
frac_contained = float(isect.area) / float(county_shape.area)
pop['frac_contained'] = frac_contained
#print("contained: {}".format(frac_contained))
else:
shp['intersection'] = shp['geometry']
shp['union'] = shp['geometry']
pop['area'] = county_shape.area
frac_contained = 1.0
pop['frac_contained'] = 1.0
# compute effective population
#print("contained population: {}".format(frac_contained * raw_pop))
pop['effective'] = frac_contained * raw_pop
p['population'] = pop
shp['properties'] = p
# add area
compute_land_area(shp)
# have to buffer the results to see how many colors to generate
results = list(pbps.get_shapes(area, dist))
color_list = plt.cm.Dark2(np.linspace(0, 1, len(results)))
# get the shapes covered by the coverage area.
for i, shp in enumerate(results):
p = shp['properties']
#track states we hit
if p['STATE'] not in states:
states.add(p['STATE'])
# also store the goeid so we dont plot again later
geoid[p['GEO_ID']] = shp
label = "{} {} [{}]".format(state_name(p['STATE']),
county_name(p['STATE'], p['COUNTY']),
p['TRACT'])
#print ("-"*10 + label + "-"*10)
#pprint(p['population'])
ec = 'black'
lw = 2
if p['population']['effective'] < 0.01:
ec = 'blue'
lw = 4
patches = make_patch(shp['geometry'],
fc=color_list[i],
lw=lw,
ec=ec,
def plot_single_place(outputfmt, placename, cover_by_place_q):
#ps = PopulationBasedPointSampler()
connection = pymongo.MongoClient(
'mongodb://*****:*****@eg-mongodb.bucknell.edu/ym015')
db = connection.get_default_database()
e = Elevation() # to look up point elevations.
query = {'name': placename}
cover_by_place = []
lossbynumbase = {}
loss_threshold = {}
# qfilter = {'_id':1}
# if not os.path.exists(outputfmt.format(placename, i, 'elevation')):
# qfilter ['shapes'] = 1
# qfilter ['points'] = 1
# have to talk out the filter for just id to comptue other values.
cursor = db['POINTS'].find(query, {
'_id': 1,
'LSAD': 1,
'state': 1,
'points': 1,
'shapes': 1,
'shape_ids': 1
},
no_cursor_timeout=True) #.limit(2)
state = None
for i, pointdoc in enumerate(cursor):
state = pointdoc['state']
print("{}: {} of {}.".format(placename, i + 1, cursor.count()))
lqry = {'point_docid': pointdoc['_id'], 'tx_height': 5, 'rx_height': 1}
#lqry = {'point_docid': pointdoc['_id']}
if not os.path.exists(outputfmt.format(placename, i, 'loss')):
if i == 0:
for rsltdoc in db['POINTRESULTS'].find(lqry, {
'nodes': 1,
'num_basestations': 1,
'loss_threshold': 1
}):
lossbynumbase[rsltdoc['num_basestations']] = [
x['min_loss'] for x in rsltdoc['nodes']
if not math.isnan(x['min_loss'])
]
loss_threshold[rsltdoc['num_basestations']] = [
rsltdoc['loss_threshold']
]
else:
for rsltdoc in db['POINTRESULTS'].find(lqry, {
'nodes': 1,
'num_basestations': 1,
'loss_threshold': 1
}):
lossbynumbase[rsltdoc['num_basestations']] += [
x['min_loss'] for x in rsltdoc['nodes']
if not math.isnan(x['min_loss'])
]
loss_threshold[rsltdoc['num_basestations']].append(
rsltdoc['loss_threshold'])
#cqry = {'point_docid': pointdoc['_id'], 'tx_height': 5, 'rx_height': 1, 'num_basestations': 1}
cqry = {'point_docid': pointdoc['_id'], 'num_basestations': 1}
cover_by_place += [
x['connected']
for x in db['POINTRESULTS'].find(cqry, {'connected': 1})
]
#skip the rest, for now.
#continue
#print(lossbynumbase)
#print(loss_threshold)
# stats are the same, so only do it once.
if i == 0:
if (not os.path.exists(outputfmt.format(placename, i, 'elevation'))) or \
(not os.path.exists(outputfmt.format(placename, i, 'stats'))):
fig = plt.figure(figsize=(8, 8))
ax = plt.subplot(111)
if pointdoc['LSAD'] == 'County':
c_shp_query = {
'properties.NAME': placename,
'properties.STATE': pointdoc['state']
}
c_shp = list(db['GENZ2010_050'].find(c_shp_query))
#plot_shapes(ax, city_shp, filled= True, show_states = True, fc=lambda x:'red', alpha=0.99)
plot_shapes(ax, c_shp, filled=False, show_states=True)
elif pointdoc['LSAD'] == 'city':
city_shp_query = {
'properties.NAME': placename,
'properties.STATE': pointdoc['state']
}
city_shp = list(db['GENZ2010_160'].find(city_shp_query))
#plot_shapes(ax, city_shp, filled= True, show_states = True, fc=lambda x:'red', alpha=0.99)
plot_shapes(ax, city_shp, filled=False, show_states=True)
else:
print(
"WARN: unknown LSAD: {}, not plotting outline".format(
pointdoc['LSAD']))
if 'shapes' not in pointdoc:
print("getting shapes...")
sqry = {'_id': {'$in': pointdoc['shape_ids']}}
shapes = list(db['GENZ2010_140'].find(sqry))
else:
shapes = pointdoc['shapes']
plot_shapes(ax, shapes, filled=True, show_states=True)
# this is one way to do it
# tract_shp = list(ps.get_tract_shapes_in_area(city_shp[0]))
# plot_shapes(ax, tract_shp, filled= True, show_states = True)
elev_data = []
for pt in pointdoc['points']:
ax.plot(pt['coordinates'][0],
pt['coordinates'][1],
'.',
color='black',
ms=4,
lw=2,
alpha=0.75)
elev_data.append(e.lookup(pt['coordinates']))
ax.ticklabel_format(useOffset=False, style="plain")
ax.set_title("{}, {}".format(placename, state_name(state)))
ax.axis('equal')
plt.tight_layout()
plt.savefig(outputfmt.format(placename, i, 'tracts'))
plt.close(fig)
# show elevation distribution
fig = plt.figure(figsize=(8, 8))
ax = plt.subplot(111)
ax.hist(elev_data,
normed=1,
fill=False,
ec="black",
lw=3,
hatch=".")
ax.set_title(placename)
ax.set_xlabel('Elevation (meters)')
ax.set_ylabel('Rate')
ax.axis('tight')
plt.savefig(outputfmt.format(placename, i, 'elevation'))
plt.close(fig)
print("{} has {} tracts.".format(placename, len(shapes)))
areas = [
sh['properties']['area']['effective'] / (1000.0**2)
for sh in shapes if 'area' in sh['properties']
] # convert meters to KM
pops = [
sh['properties']['population']['effective'] / 1000.0
for sh in shapes if 'area' in sh['properties']
]
#print (areas)
#print (pops)
print('total pop: ', sum(pops), 'median: ', numpy.median(pops))
print('total area: ', sum(areas), 'median: ',
numpy.median(areas))
if not os.path.exists(outputfmt.format(placename, i, 'stats')):
if areas != None and pops != None:
fig = plt.figure(figsize=(8, 4))
ax = plt.subplot(121)
ax.hist(areas,
normed=0,
fill=False,
ec='black',
lw=3,
hatch='/')
ax.set_title("Area")
ax.set_xlabel('Square Kilometers')
ax.set_ylabel('Count')
ax.axis('normal')
ax = plt.subplot(122)
ax.set_title("Population")
ax.hist(pops,
normed=0,
fill=False,
ec='black',
lw=3,
hatch='x')
ax.set_xlabel('Population (Thousands)')
ax.set_ylabel('Count')
ax.axis('normal')
#ax.violinplot ([areas, pops], showmeans=False, showmedians=True)
#ax.set_xticklabels(['Area', 'Population'])
plt.tight_layout()
plt.savefig(outputfmt.format(placename, i, 'stats'))
plt.close(fig)
#if not os.path.exists(outputfmt.format(placename, i, 'loss')):
if 1:
# show loss distribution
fig = plt.figure(figsize=(8, 4))
ax = plt.subplot(111)
bases = sorted(lossbynumbase.keys())
if len(bases) > 0:
# print ('plot bases: ', end="")
# pprint(bases)
# print ('loss values: ', end="")
# pprint ([lossbynumbase[bs] for bs in bases])
ax.violinplot([lossbynumbase[bs] for bs in bases],
bases,
showmeans=False,
showmedians=True)
# print('medians are: ', end="")
# pprint([numpy.median(loss_threshold[i]) for i in bases])
ax.plot(bases, [numpy.median(loss_threshold[i]) for i in bases],
'--',
alpha=0.5,
color='green',
lw=2)
ax.set_title("{}, {}".format(placename, state_name(state)))
ax.set_xlabel('Number of basestations')
ax.set_ylabel('Loss (dB)')
ax.axis('tight')
plt.tight_layout()
plt.savefig(outputfmt.format(placename, i, 'loss'))
plt.close(fig)
else:
print(lossbynumbase)
print(bases)
cover_by_place_q.put((placename, cover_by_place))
