def set_grid(self, lat0, lon0, width, length, grid_size=0.05):
"""
Set Target area based on given centroid coordinate and w, l of the rectangle area
Input:
lat0, lon0: Centroid coordinate
width, length: Width, Length of the area
grid_size: width of each grid, in km
Post-Condition:
self.grids updated
"""
# Generate grid coordinate in (x, y)
xs = np.arange(float(-width + grid_size)/2, float(width + grid_size)/2, grid_size)
ys = np.arange(float(-length + grid_size)/2, float(length + grid_size)/2, grid_size)
# Convert to coordinate
index = 1
start = geopy.Point(lat0, lon0)
for _x in xs:
for _y in ys:
# Get Coordinate of (_x, _y) with (lat0, lon0) at (0, 0)
north = gd.VincentyDistance(kilometers=_y)
east = gd.VincentyDistance(kilometers=_x)
point = north.destination(east.destination(start, 90), 0)
self.grids.update({index: (point[0], point[1])})
# Update index
index += 1
def get_gains():
"""Returns lat and lon gain
Gain is space between circles.
"""
start = Point(*get_map_center())
base = config.SCAN_RADIUS * math.sqrt(3)
height = base * math.sqrt(3) / 2
dis_a = distance.VincentyDistance(meters=base)
dis_h = distance.VincentyDistance(meters=height)
lon_gain = dis_a.destination(point=start, bearing=90).longitude
lat_gain = dis_h.destination(point=start, bearing=0).latitude
return abs(start.latitude - lat_gain), abs(start.longitude - lon_gain)
def getPoints(lat0, lon0, x, y):
"""
Convert x,y (in km) to lat,long with given (lat, lon) represents (0,0) in km
Inputs:
lat0, lon0: coordinates of (0, 0) in km
x, y (iterables such as list or ndarray): coordinates of points in km.
Returns:
the coordinate(s) in lat, lon
"""
start = geopy.Point(lat0, lon0)
points = []
for _x, _y in zip(x, y):
north = gd.VincentyDistance(kilometers=_y)
east = gd.VincentyDistance(kilometers=_x)
points.append(north.destination(east.destination(start, 90), 0))
lats, lons = zip(*points)[:2]
return lats, lons
def get_queryset(self):
if not self.request.query_params:
return Party.objects.all()
else:
user = self.request.user.id
lat = self.request.query_params.get('lat')
lng = self.request.query_params.get('lng')
if lat is None or lng is None:
return Party.objects.none()
d = self.request.query_params.get('d')
if d is None:
d = 10.0
d = convert_mile_to_kilometer(float(d))
origin = Point(lat, lng)
north = distance.VincentyDistance(kilometers=d).destination(
origin, 0)
east = distance.VincentyDistance(kilometers=d).destination(
origin, 90)
south = distance.VincentyDistance(kilometers=d).destination(
origin, 180)
west = distance.VincentyDistance(kilometers=d).destination(
origin, 270)
north_lat = north.latitude
east_lng = east.longitude
south_lat = south.latitude
west_lng = west.longitude
q = Party.objects\
.filter(lat__gte=south_lat, lat__lte=north_lat)\
.exclude(host__id=user).exclude(bouncers__id=user).exclude(invitees__id=user)
# edge case around the anti-meridian
if west_lng > east_lng:
# it seems like you can't chain Q-object queries with regular queries
return q.filter(Q(lng__gte=west_lng) | Q(lng__lte=east_lng))
else:
return q.filter(lng__gte=west_lng, lng__lte=east_lng)
def getAngle(angleAttempts, fromA, toB):
start = Point(toB['lat'], toB['lng'])
travelDistance = gDistance.VincentyDistance(kilometers=0.01)
def worker(acc, bearing):
destination = travelDistance.destination(point=start, bearing=bearing)
distancePoint = gDistance.distance(
Point(fromA['lat'], fromA['lng']),
Point(destination.latitude, destination.longitude)).km
if (distancePoint > acc['distance']):
return {'bearing': bearing, 'distance': distancePoint}
return acc
r = reduce(worker, angleAttempts, {'bearing': -1, 'distance': -1})
return r['bearing']
def get_bounding_box_of_circle(center_lon, center_lat, radious):
start = geopy.Point(latitude=center_lat, longitude=center_lon)
d = distance.VincentyDistance(kilometers=radious)
# bearing=0 -> north
# bearing=90 -> east
# bearing=180 -> south
# bearing=270 -> west
east_north_point = {
'longitude': d.destination(point=start, bearing=90).longitude,
'latitude': d.destination(point=start, bearing=0).latitude
}
west_south_point = {
'longitude': d.destination(point=start, bearing=270).longitude,
'latitude': d.destination(point=start, bearing=180).latitude
}
return west_south_point, east_north_point
def get_coordinates(start_point, end_point, distance_meters):
"""
Calculates the new coordinates between two points depending
of the specified distance and the calculated bearing.
Parameters
----------
start_point: geopy.Point
end_point: geopy.Point
distance_meters: float
Returns
-------
point: geopy.Point
A new point between the start and the end points.
"""
bearing = get_bearing(start_point, end_point)
distance_km = distance_meters / 1000
d = geo_dist.VincentyDistance(kilometers=distance_km)
destination = d.destination(point=start_point, bearing=bearing)
return geopy.Point(destination.latitude, destination.longitude)
def get_nearest(self, num=5):
"Returns the nearest X people, but only within the same continent"
# TODO: Add caching
people = list(self.country.djangoperson_set.select_related().exclude(
pk=self.id,
))
if len(people) <= num:
# Not enough in country
# use people from the same continent instead
people = list(DjangoPerson.objects.filter(
country__continent=self.country.continent,
).exclude(pk=self.id).select_related())
# Sort and annotate people by distance
for person in people:
person.distance_in_miles = distance.VincentyDistance(
(self.latitude, self.longitude),
(person.latitude, person.longitude)
).miles
# Return the nearest X
people.sort(key=lambda x: x.distance_in_miles)
return people[:num]
filename = os.path.expanduser('~/.cache/guifinetstudio/detail/26494.cnml')
nid = 33968 # MLGInvisible
if len(sys.argv) == 3:
nid, filename = sys.argv[1:2]
# Parse CNML file
cnmlp = CNMLParser(filename)
from_node = cnmlp.getNode(nid)
from_coord = (from_node.latitude, from_node.longitude)
nodes = cnmlp.getNodes()
distances = []
nodes.remove(from_node)
# Calculate distance to every node in the same zone
for to_node in nodes:
to_coord = (to_node.latitude, to_node.longitude)
dist = distance.VincentyDistance(from_coord, to_coord)
bearing = calcBearing(from_node.latitude, from_node.longitude,
to_node.latitude, to_node.longitude)
distances.append((dist.km, from_node.title, to_node.title, bearing))
# Sort by distance
for d in sorted(distances):
if d[0] >= 1:
dist_metric = '%.3f Km' % d[0]
else:
dist_metric = '%d m' % (d[0] * 1000)
print('{} -> {}: {} (Az: {:.2f})'.format(d[1], d[2], dist_metric,
d[3]))
train = df.query('Sold < @d')
test = df.query('Sold == @d')
actual += list(test['PRICE'])
# Model 1 : Full Variance
ybar = train['lPRICE'].mean()
model1 += [ybar] * len(test.index)
# Model 2 : Simple OLS
m2 = LinearRegression()
m2.fit(train[pcols], train[['lPRICE']])
ypred = m2.predict(test[pcols])
model2 += list(ypred.ravel())
for i in list(test.index):
tgt_lat = test['LATITUDE'][i]
tgt_lon = test['LONGITUDE'][i]
c1 = (tgt_lat, tgt_lon)
train['Dist'] = map(lambda x, y: gd.VincentyDistance(c1, (x, y)).miles,
train['LATITUDE'], train['LONGITUDE'])
train['dWgt'] = 1 / train['Dist']
train['dWgt'] = np.where(train['dWgt'] > 25, 25, train['dWgt'])
train['Wgt'] = train['tWgt'] * train['dWgt']
# Model 3 : Weighted OLS
m3 = LinearRegression()
m3.fit(train[pcols], train[['lPRICE']], np.array(train['Wgt']))
ypred = m3.predict(test[pcols].query('index == @i'))
model3 += list(ypred.ravel())
# Model 4 : Weighted SVR
# TODO: need to standardize the data, then use SVR
m4 = SVR(C=0.1)
m4.fit(train[pcols], train[['lPRICE']], np.array(train['Wgt']))
ypred = m4.predict(test[pcols].query('index == @i'))
