def _getTimeSeries(self):
if self.data_source == 'instagram':
its = InstagramTimeSeries(self.region, self.data_backward, self.current_time)
elif self.data_source == 'twitter':
pass
ts = its.buildTimeSeries()
return ts
def run():
coordinates = [InstagramConfig.photo_min_lat,
InstagramConfig.photo_min_lng,
InstagramConfig.photo_max_lat,
InstagramConfig.photo_max_lng
]
huge_region = Region(coordinates)
alarm_region_size = 25
regions = huge_region.divideRegions(alarm_region_size,alarm_region_size)
filtered_regions = huge_region.filterRegions( regions)
regions = filtered_regions
test_cnt = 0
print 'all regions',len(regions)
for region in regions:
#delete the last 7*24*3600 to set it back to Dec 1st
start_of_time = 1354320000 #+ 7*24*3600
end_of_time = 1354320000 + 7*24*3600 #+ 7*24*3600
series = InstagramTimeSeries( region, start_of_time, end_of_time)
series = series.buildTimeSeries()
region.display()
for t in series.index:
print t,',',series[t]
print '\n'
def run():
coordinates = [
InstagramConfig.photo_min_lat, InstagramConfig.photo_min_lng,
InstagramConfig.photo_max_lat, InstagramConfig.photo_max_lng
]
huge_region = Region(coordinates)
alarm_region_size = 25
regions = huge_region.divideRegions(alarm_region_size, alarm_region_size)
filtered_regions = huge_region.filterRegions(regions)
regions = filtered_regions
test_cnt = 0
print 'all regions', len(regions)
for region in regions:
#delete the last 7*24*3600 to set it back to Dec 1st
start_of_time = 1354320000 #+ 7*24*3600
end_of_time = 1354320000 + 7 * 24 * 3600 #+ 7*24*3600
series = InstagramTimeSeries(region, start_of_time, end_of_time)
series = series.buildTimeSeries()
region.display()
for t in series.index:
print t, ',', series[t]
print '\n'
def _computeVariation(self):
values = [0]*24
ts = InstagramTimeSeries(self.region, self.training_start_time, self.training_end_time)
instagram_ts = ts.buildTimeSeries()
M = np.zeros([1000,24])
initial_date = instagram_ts.index[0]
for idx in instagram_ts.index:
if not pd.isnull( instagram_ts[idx] ):
day_dif = (idx - initial_date).days
M[day_dif, idx.hour] = instagram_ts[idx]
max_day = (instagram_ts.index[len(instagram_ts)-1] - initial_date ).days
M = M[0:max_day, :]
return np.mean(M, axis=0), np.std(M,axis=0)
def _computeVariation(self):
values = [0] * 24
ts = InstagramTimeSeries(self.region, self.training_start_time,
self.training_end_time)
instagram_ts = ts.buildTimeSeries()
M = np.zeros([1000, 24])
initial_date = instagram_ts.index[0]
for idx in instagram_ts.index:
if not pd.isnull(instagram_ts[idx]):
day_dif = (idx - initial_date).days
M[day_dif, idx.hour] = instagram_ts[idx]
max_day = (instagram_ts.index[len(instagram_ts) - 1] -
initial_date).days
M = M[0:max_day, :]
return np.mean(M, axis=0), np.std(M, axis=0)
