def path_json(vc):
db = msc.mysqlcon()
cities = []
vc = vc.split(',')
for vci in vc:
que = "select nid, name, lat, lng, url, photo from foursquare join foursquare_id on foursquare.id=foursquare_id.id where nid=%s" % vci
db.query(que)
query_results = db.store_result().fetch_row(maxrows=1)
#print query_results
for result in query_results:
nname = result[1].encode('ascii', 'ignore')
cities.append(
dict(City=unicode(str(result[0]), 'utf8'),
CountryCode=nname,
lat=result[2],
lng=result[3],
url=result[4],
photo=result[5]))
db.close()
return jsonify(dict(cities=cities))
def suggest(vin, origin, destination):
conn = msc.mysqlcon()
x = conn.cursor()
# -----------
mcn = np.loadtxt('mcn.mat')
dist = np.loadtxt('dx_mat.mat')
#print vin
if vin:
vin = vin.split(',')
vin = map(lambda x: int(x), vin)
#top5, dxlist = top5dist(mcn,dist,vin,origin,destination)
top5, dxlist = top5dist(mcn, dist, vin, origin, destination, -1, 6)
#top5s = str(top5[0]) + ', ' + str(top5[1]) + ', ' + str(top5[2]) + ', ' + str(top5[3]) + ', ' + str(top5[4])
top5s = str(top5[0]) + ', ' + str(top5[1]) + ', ' + str(
top5[2]) + ', ' + str(top5[3]) + ', ' + str(top5[4]) + ', ' + str(
top5[5]) # now top 6
top5s = 'select nid, lat, lng, name, url, photo from foursquare join foursquare_id on foursquare.id=foursquare_id.id where nid in (%s) ORDER BY FIND_IN_SET(nid,"%s")' % (
top5s, top5s)
x.execute(top5s)
result = x.fetchall() # need to sort by request order
sites = []
for res in result:
# hours and minutes out of the way
sf = 2 # scale factor
dxin = dxlist[res[0]] * sf + 5000
dxihm = '%02d:%02d' % (int(dxin / 96500), int(dxin / 96500 % 1 * 60))
# distance out of the way (miles)
#sf = 1
#dxihm = round( float(dxlist[res[0]])*sf/1000/0.621371,2)
#dxihm = float(dxlist[res[0]]/1000)
sites.append(
dict(City=unicode(str(res[0]), 'utf8'),
CountryCode=res[3],
dxi=dxihm,
lat=res[1],
lng=res[2],
url=res[4],
photo=res[5])) # nid, name, time off in HHMM
try:
x.close()
conn.close()
except:
print "nothing to close"
return sites
def make_dist_vec(s1,e1,M=-1):
# makes the distance vector given start and end point
# passing in lat,lng coordinates
s1latlng = s1.split(',')
s1latlng = [float(s1latlng[0]),float(s1latlng[1])]
e1latlng = e1.split(',')
e1latlng = [float(e1latlng[0]),float(e1latlng[1])]
#print e1latlng
if M==-1:
conn = msc.mysqlcon()
x = conn.cursor()
x.execute('select nid, lat, lng from foursquare join foursquare_id on foursquare.id=foursquare_id.id order by nid');
result = x.fetchall()
lat_s = s1latlng[0]
lng_s = s1latlng[1]
lat_e = e1latlng[0]
lng_e = e1latlng[1]
dist_s = [distance_on_unit_sphere(lat_s,lng_s,lat_e,lng_e)]
dist_e = [dist_s[0]]
for c1 in range(len(result)):
lat1 = result[c1][1]
lng1 = result[c1][2]
dr_s = distance_on_unit_sphere(lat1, lng1, lat_s, lng_s)
dr_e = distance_on_unit_sphere(lat1, lng1, lat_e, lng_e)
dist_s.append(dr_s)
dist_e.append(dr_e)
if M==-1:
conn.close()
x.close()
return dist_s, dist_e
def make_dist_vec(s1, e1, M=-1):
# makes the distance vector given start and end point
# passing in lat,lng coordinates
s1latlng = s1.split(',')
s1latlng = [float(s1latlng[0]), float(s1latlng[1])]
e1latlng = e1.split(',')
e1latlng = [float(e1latlng[0]), float(e1latlng[1])]
#print e1latlng
if M == -1:
conn = msc.mysqlcon()
x = conn.cursor()
x.execute(
'select nid, lat, lng from foursquare join foursquare_id on foursquare.id=foursquare_id.id order by nid'
)
result = x.fetchall()
lat_s = s1latlng[0]
lng_s = s1latlng[1]
lat_e = e1latlng[0]
lng_e = e1latlng[1]
dist_s = [distance_on_unit_sphere(lat_s, lng_s, lat_e, lng_e)]
dist_e = [dist_s[0]]
for c1 in range(len(result)):
lat1 = result[c1][1]
lng1 = result[c1][2]
dr_s = distance_on_unit_sphere(lat1, lng1, lat_s, lng_s)
dr_e = distance_on_unit_sphere(lat1, lng1, lat_e, lng_e)
dist_s.append(dr_s)
dist_e.append(dr_e)
if M == -1:
conn.close()
x.close()
return dist_s, dist_e
def suggest(vin,origin,destination):
conn = msc.mysqlcon()
x = conn.cursor()
# -----------
mcn=np.loadtxt('mcn.mat')
dist=np.loadtxt('dx_mat.mat')
#print vin
if vin:
vin = vin.split(',')
vin = map(lambda x: int(x), vin)
#top5, dxlist = top5dist(mcn,dist,vin,origin,destination)
top5, dxlist = top5dist(mcn,dist,vin,origin,destination,-1,6)
#top5s = str(top5[0]) + ', ' + str(top5[1]) + ', ' + str(top5[2]) + ', ' + str(top5[3]) + ', ' + str(top5[4])
top5s = str(top5[0]) + ', ' + str(top5[1]) + ', ' + str(top5[2]) + ', ' + str(top5[3]) + ', ' + str(top5[4]) + ', ' + str(top5[5]) # now top 6
top5s = 'select nid, lat, lng, name, url, photo from foursquare join foursquare_id on foursquare.id=foursquare_id.id where nid in (%s) ORDER BY FIND_IN_SET(nid,"%s")' % (top5s,top5s)
x.execute(top5s);
result = x.fetchall() # need to sort by request order
sites = []
for res in result:
# hours and minutes out of the way
sf = 2 # scale factor
dxin = dxlist[res[0]]*sf+5000
dxihm = '%02d:%02d' % (int(dxin/96500),int(dxin/96500 % 1 * 60))
# distance out of the way (miles)
#sf = 1
#dxihm = round( float(dxlist[res[0]])*sf/1000/0.621371,2)
#dxihm = float(dxlist[res[0]]/1000)
sites.append(dict(City=unicode(str(res[0]), 'utf8'), CountryCode=res[3], dxi=dxihm, lat=res[1], lng=res[2], url=res[4], photo=res[5] )) # nid, name, time off in HHMM
try:
x.close()
conn.close()
except:
print "nothing to close"
return sites
def path_json(vc):
db = msc.mysqlcon()
cities = []
vc = vc.split(',')
for vci in vc:
que = "select nid, name, lat, lng, url, photo from foursquare join foursquare_id on foursquare.id=foursquare_id.id where nid=%s" % vci
db.query(que)
query_results = db.store_result().fetch_row(maxrows=1)
#print query_results
for result in query_results:
nname = result[1].encode('ascii','ignore')
cities.append( dict(City=unicode(str(result[0]), 'utf8'), CountryCode=nname, lat=result[2], lng=result[3], url=result[4], photo=result[5]) )
db.close()
return jsonify(dict(cities=cities))
def top5dist(mcn,dist,rrest,start1='Palo Alto, CA',end1='Stanford, CA',M=-1,nc=5):
sugg = mcn[0]-mcn[0]
for r in rrest:
sugg += mcn[r]
for r in rrest:
sugg[r]=0
# make NxM matrix of relevant distances
dmat = []
for rr in rrest:
dmat.append(np.append(0,dist[rr-1]))
# append start and end locations
dist_start, dist_end = make_dist_vec(start1,end1)
dmat.append(dist_start)
dmat.append(dist_end)
dmat = np.array(dmat)
# find closest points
dx = [0]
for ig in range(len(mcn)):
if ig:
closest = dmat[:,ig]
cig = closest.argsort()[:2]
x1 = closest[cig[0]]+closest[cig[1]]
# insert if cases here
if (len(rrest) in cig) and ((len(rrest)+1) in cig):
x2 = dmat[len(rrest)][0]
elif (len(rrest) in cig):
x2 = dmat[len(rrest)][rrest[min(cig)]]
elif ((len(rrest)+1) in cig):
x2 = dmat[len(rrest)+1][rrest[min(cig)]]
else:
x2 = dist[rrest[cig[0]]-1][rrest[cig[1]]-1] # shift by 1
dx.append(abs(x1-x2))
# combine sugg with distance
ranks = []
for ig in range(len(sugg)):
ranks.append(sugg[ig]*(np.exp(-dx[ig]**2/(2*5643.81**2))+0.00735612)) # fit from data
ranks = np.array(ranks)
# if not enough suggestions fill in by tips
nc0 = sum(map(lambda x: x>0, ranks))
nc1 = 0
if ( nc0 < nc):
nc1 = nc - sum(map(lambda x: x>0, ranks))
if (M==-1):
conn = msc.mysqlcon()
x = conn.cursor()
x.execute('select good, tips from foursquare join foursquare_id on foursquare.id=foursquare_id.id order by nid');
result = x.fetchall()
#print 'res list ' + str(len(result))
toplist_supp = [0]
for jg in range(len(result)):
if (jg+1 in rrest):
tipxdist = 0 # no repeats
else:
tipxdist = result[jg][0]*float(result[jg][1])*(np.exp(-dx[jg+1]**2/(2*5643.81**2))+0.00735612)
toplist_supp.append(tipxdist)
# combine the two top recommendation lists
toplist = ranks.argsort()[-nc:][::-1]
if ( nc0 < nc):
toplist_supp = np.array(toplist_supp).argsort()[-nc:][::-1]
if (M==-1):
conn.close()
x.close()
#print toplist
top5list = []
for ig in range(min(nc0,nc)):
top5list.append(toplist[ig])
for ig in range(nc1):
top5list.append(toplist_supp[ig])
return top5list, dx
import os
import sys
import MySQLdb
import itertools as itt
import numpy as np
import mysqlconnect as msc
conn = msc.mysqlcon()
x = conn.cursor()
x.execute('select max(region) from instagrams_mult');
mcsize = x.fetchall()
mcsize = mcsize[0][0]+1
# make NxN matrix
mc = [0 for ro in xrange(mcsize)]
mc = [mc[:] for co in xrange(mcsize)]
x.execute('select distinct uid from instagrams_mult')
result = x.fetchall()
lenres = len(result)
ni = 0
for res in result:
ni+=1
if not ni%1000:
print str(ni) + '/' + str(lenres)
x.execute('select distinct region from instagrams_mult where uid=%s',(res[0],))
regions = x.fetchall()
import os
import sys
import MySQLdb
import itertools as itt
import numpy as np
import mysqlconnect as msc
conn = msc.mysqlcon()
x = conn.cursor()
x.execute('select max(region) from instagrams_mult')
mcsize = x.fetchall()
mcsize = mcsize[0][0] + 1
# make NxN matrix
mc = [0 for ro in xrange(mcsize)]
mc = [mc[:] for co in xrange(mcsize)]
x.execute('select distinct uid from instagrams_mult')
result = x.fetchall()
lenres = len(result)
ni = 0
for res in result:
ni += 1
if not ni % 1000:
print str(ni) + '/' + str(lenres)
x.execute('select distinct region from instagrams_mult where uid=%s',
(res[0], ))
regions = x.fetchall()
def top5dist(mcn,
dist,
rrest,
start1='Palo Alto, CA',
end1='Stanford, CA',
M=-1,
nc=5):
sugg = mcn[0] - mcn[0]
for r in rrest:
sugg += mcn[r]
for r in rrest:
sugg[r] = 0
# make NxM matrix of relevant distances
dmat = []
for rr in rrest:
dmat.append(np.append(0, dist[rr - 1]))
# append start and end locations
dist_start, dist_end = make_dist_vec(start1, end1)
dmat.append(dist_start)
dmat.append(dist_end)
dmat = np.array(dmat)
# find closest points
dx = [0]
for ig in range(len(mcn)):
if ig:
closest = dmat[:, ig]
cig = closest.argsort()[:2]
x1 = closest[cig[0]] + closest[cig[1]]
# insert if cases here
if (len(rrest) in cig) and ((len(rrest) + 1) in cig):
x2 = dmat[len(rrest)][0]
elif (len(rrest) in cig):
x2 = dmat[len(rrest)][rrest[min(cig)]]
elif ((len(rrest) + 1) in cig):
x2 = dmat[len(rrest) + 1][rrest[min(cig)]]
else:
x2 = dist[rrest[cig[0]] - 1][rrest[cig[1]] - 1] # shift by 1
dx.append(abs(x1 - x2))
# combine sugg with distance
ranks = []
for ig in range(len(sugg)):
ranks.append(sugg[ig] *
(np.exp(-dx[ig]**2 /
(2 * 5643.81**2)) + 0.00735612)) # fit from data
ranks = np.array(ranks)
# if not enough suggestions fill in by tips
nc0 = sum(map(lambda x: x > 0, ranks))
nc1 = 0
if (nc0 < nc):
nc1 = nc - sum(map(lambda x: x > 0, ranks))
if (M == -1):
conn = msc.mysqlcon()
x = conn.cursor()
x.execute(
'select good, tips from foursquare join foursquare_id on foursquare.id=foursquare_id.id order by nid'
)
result = x.fetchall()
#print 'res list ' + str(len(result))
toplist_supp = [0]
for jg in range(len(result)):
if (jg + 1 in rrest):
tipxdist = 0 # no repeats
else:
tipxdist = result[jg][0] * float(
result[jg][1]) * (np.exp(-dx[jg + 1]**2 /
(2 * 5643.81**2)) + 0.00735612)
toplist_supp.append(tipxdist)
# combine the two top recommendation lists
toplist = ranks.argsort()[-nc:][::-1]
if (nc0 < nc):
toplist_supp = np.array(toplist_supp).argsort()[-nc:][::-1]
if (M == -1):
conn.close()
x.close()
#print toplist
top5list = []
for ig in range(min(nc0, nc)):
top5list.append(toplist[ig])
for ig in range(nc1):
top5list.append(toplist_supp[ig])
return top5list, dx
def open_mysql(): global conn global x conn = msc.mysqlcon() x = conn.cursor()
def open_mysql():
global conn
global x
conn = msc.mysqlcon()
x = conn.cursor()
