mean_daily_variance = NO2.groupby(lambda x: (x.date)).std()
for row in mean_daily_variance.index:
if mean_daily_variance.mean(1)[row] < 1:
valid_dates = NO2.index[NO2.index.date != row]
NO2 = NO2.transpose()[valid_dates].transpose()
#ffill and bfill NA's
NO2 = NO2.ffill().bfill()
distance = pandas.DataFrame()
for rack in bikes_metadata:
distance[rack] = pandas.Series(dict((k, opensensor.getDistance(v,bikes_metadata[rack])) for k, v in LAQN_locations.items()))
import scipy.stats as stats
distance = distance.transpose()[list(NO2.columns)].transpose()
#work out the weight for each LAQN sensor
#2 is arbitrary decay rate
weight = np.exp(-2*distance)/(np.exp(-2*distance)).sum()
rack_AQ = pandas.DataFrame()
for rack in weight.columns:
rack_AQ[rack] = (NO2 * weight[rack]).sum(axis = 1)
#set the decay parameters for the exp moving avgs
nslow = 6
#the last date defaults to today
end_date =datetime.datetime.now().strftime("%Y%m%d")
#Get the meta data not on opensensors at the moment
with open(LAQN_file) as f:
reader = csv.reader(f)
LAQN_locations = dict((rows[0],[float(rows[1]),float(rows[2])]) for rows in reader)
#create instance of Air Quality Object
opensensor = London(API_KEY)
#set my postcode
myPostCode = 'E1W 2PA'
#get the geo data for my postcode
geo_meta_data=opensensor.getPostcode(myPostCode)
#get the distance from all the LAQN sensors from my postcode
distance = pandas.Series(dict((k, opensensor.getDistance(v,[geo_meta_data['geo']['lng'], geo_meta_data['geo']['lat']])) for k, v in LAQN_locations.items()))
#work out the weight for each LAQN sensor
weight = np.exp(-2*distance)/sum(np.exp(-2*distance))
distance = pandas.concat([distance, weight], axis=1)
distance.columns = ['distance','weight']
#get the list of LAQN measures
LAQN_mtx = opensensor.getAirQuality(start=start_date,end=end_date)
#grab the NO2 data where we bhave the intersect of weights and NO2
NO2 = LAQN_mtx['NO2'][list(weight[list(LAQN_mtx['NO2'].columns)].index)]
NO2_weight = weight[list(NO2.columns)]/weight[list(NO2.columns)].sum()
#sort the matrices
NO2 = NO2.reindex_axis(sorted(NO2.columns), axis=1)
NO2_weight = NO2_weight.reindex(sorted(NO2_weight.index))
