In [107]:

    

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline

sns.set(color_codes=True)


    

In [250]:

    

data = pd.read_csv('../data/belorusskoye_in_out_1_min_2016_04.txt', delimiter='\t')
data.head()


    

    
Out[250]:






  

    

      

      
station_code
      
station_name
      
month
      
week_day
      
day_type
      
hour
      
minute
      
passengers_in
      
passenger_out
    
  
  

    

      
0
      
2000006
      
МОСКВА БЕЛОРУССКАЯ
      
апрель
      
понедельник
      
раб
      
0
      
0
      
11
      
2
    
    

      
1
      
2000006
      
МОСКВА БЕЛОРУССКАЯ
      
апрель
      
понедельник
      
раб
      
0
      
1
      
8
      
0
    
    

      
2
      
2000006
      
МОСКВА БЕЛОРУССКАЯ
      
апрель
      
понедельник
      
раб
      
0
      
2
      
7
      
0
    
    

      
3
      
2000006
      
МОСКВА БЕЛОРУССКАЯ
      
апрель
      
понедельник
      
раб
      
0
      
3
      
14
      
2
    
    

      
4
      
2000006
      
МОСКВА БЕЛОРУССКАЯ
      
апрель
      
понедельник
      
раб
      
0
      
4
      
11
      
2
    
  

In [ ]:

    




    

In [98]:

    




    

In [251]:

    

data_wed = data.query('week_day == "wed"')
stations = map(lambda x: x, set(data['station_name'].as_matrix()))


    

In [253]:

    

pass_in = list(0 for i in range(1440))
pass_out = list(0 for i in range(1440))


    

In [254]:

    

pd.DataFrame(stations)


    

    
Out[254]:






  

    

      

      
0
    
  
  

    

      
0
      
СЕТУНЬ
    
    

      
1
      
КУНЦЕВО 1
    
    

      
2
      
ФИЛИ
    
    

      
3
      
МОЖАЙСК
    
    

      
4
      
ТУЧКОВО
    
    

      
5
      
БЕГОВАЯ
    
    

      
6
      
БАКОВКА
    
    

      
7
      
ГОЛИЦЫНО
    
    

      
8
      
ЖАВОРОНКИ
    
    

      
9
      
ОТРАДНОЕ СМОЛ.
    
    

      
10
      
ОДИНЦОВО
    
    

      
11
      
МОСКВА БЕЛОРУССКАЯ
    
  

In [255]:

    

belorus = pd.DataFrame(data.query('station_name == "МОСКВА БЕЛОРУССКАЯ"'))


    

In [256]:

    

belorus['time'] = 60 * belorus['hour'] + belorus['minute']


    

In [257]:

    

window(belorus['passengers_in'].as_matrix().transpose()).shape


    

    
Out[257]:





(8630,)

In [258]:

    

belorus['time'][N-1:].shape


    

    
Out[258]:





(8630,)

In [269]:

    

N = 30
def window(x):
    return np.convolve(x, np.ones((N,))/N, mode='valid')

plt.plot(belorus['time'][N-1:], window(belorus['passengers_in'].as_matrix().transpose()), 'r-')
plt.plot(belorus['time'][N-1:], window(belorus['passenger_out'].as_matrix().transpose()), 'b-')
plt.axis([0,1440,0,600])
plt.show()


    

In [261]:

    

def in_out_dict(f):
    return [{'time' : row[1]['time'], 'in': row[1]['passengers_in'], 'out': row[1]['passenger_out']} 
        for row in f.iterrows()]


    

In [262]:

    

from json import dumps, dump
import os


    

In [ ]:

    




    

In [265]:

    

path = '../data/stations_in_out/'
if not os.path.exists(path):
    os.mkdir(path)
    
for station in stations:
    for day_type in set(data['day_type'].values):
        for week_day in set(data['week_day'].values):
            station_data = pd.DataFrame(data.query('station_name == "{0}" and week_day == "{1}" and day_type == "{2}"'.format(station, week_day, day_type)))
            station_data['time'] = 60 * station_data['hour'] + station_data['minute']
            result = in_out_dict(station_data)
            if not len(result) == 0:
                with open(path + 'data_{0}_{1}_{2}.json'.format(station, week_day, day_type), 'w') as f:
                    dump(in_out_dict(station_data), f)


    

In [277]:

    

for station in stations:
    for day_type in set(data['day_type'].values):
        for week_day in set(data['week_day'].values):
            station_data = pd.DataFrame(data.query('station_name == "{0}" and week_day == "{1}" and day_type == "{2}"'.format(station, week_day, day_type)))
            if len(station_data) == 0:
                continue
            station_data['time'] = 60 * station_data['hour'] + station_data['minute']
            plt.plot(station_data['time'][N-1:], window(station_data['passengers_in'].as_matrix().transpose()), 'r-')
            plt.plot(station_data['time'][N-1:], window(station_data['passenger_out'].as_matrix().transpose()), 'b-')
#             plt.axis([0,1440])
            title = 'station_name == "{0}" and week_day == "{1}" and day_type == "{2}"'.format(station, week_day, day_type).decode('utf-8')
            print title
            plt.title(title)
            plt.show()


    

    




station_name == "СЕТУНЬ" and week_day == "суббота" and day_type == "вых"






    













    




station_name == "СЕТУНЬ" and week_day == "воскресенье" and day_type == "вых"






    













    




station_name == "СЕТУНЬ" and week_day == "вторник" and day_type == "раб"






    













    




station_name == "СЕТУНЬ" and week_day == "четверг" and day_type == "раб"






    













    




station_name == "СЕТУНЬ" and week_day == "среда" and day_type == "раб"






    













    




station_name == "СЕТУНЬ" and week_day == "пятница" and day_type == "раб"






    













    




station_name == "СЕТУНЬ" and week_day == "понедельник" and day_type == "раб"






    













    




station_name == "КУНЦЕВО 1" and week_day == "суббота" and day_type == "вых"






    













    




station_name == "КУНЦЕВО 1" and week_day == "воскресенье" and day_type == "вых"






    













    




station_name == "КУНЦЕВО 1" and week_day == "вторник" and day_type == "раб"






    













    




station_name == "КУНЦЕВО 1" and week_day == "четверг" and day_type == "раб"






    













    




station_name == "КУНЦЕВО 1" and week_day == "среда" and day_type == "раб"






    













    




station_name == "КУНЦЕВО 1" and week_day == "пятница" and day_type == "раб"






    













    




station_name == "КУНЦЕВО 1" and week_day == "понедельник" and day_type == "раб"






    













    




station_name == "ФИЛИ" and week_day == "суббота" and day_type == "вых"






    













    




station_name == "ФИЛИ" and week_day == "воскресенье" and day_type == "вых"






    













    




station_name == "ФИЛИ" and week_day == "вторник" and day_type == "раб"






    













    




station_name == "ФИЛИ" and week_day == "четверг" and day_type == "раб"






    













    




station_name == "ФИЛИ" and week_day == "среда" and day_type == "раб"






    













    




station_name == "ФИЛИ" and week_day == "пятница" and day_type == "раб"






    













    




station_name == "ФИЛИ" and week_day == "понедельник" and day_type == "раб"






    













    




station_name == "МОЖАЙСК" and week_day == "суббота" and day_type == "вых"






    













    




station_name == "МОЖАЙСК" and week_day == "воскресенье" and day_type == "вых"






    













    




station_name == "МОЖАЙСК" and week_day == "вторник" and day_type == "раб"






    













    




station_name == "МОЖАЙСК" and week_day == "четверг" and day_type == "раб"






    













    




station_name == "МОЖАЙСК" and week_day == "среда" and day_type == "раб"






    













    




station_name == "МОЖАЙСК" and week_day == "пятница" and day_type == "раб"






    













    




station_name == "МОЖАЙСК" and week_day == "понедельник" and day_type == "раб"






    













    




station_name == "ТУЧКОВО" and week_day == "суббота" and day_type == "вых"






    













    




station_name == "ТУЧКОВО" and week_day == "воскресенье" and day_type == "вых"






    













    




station_name == "ТУЧКОВО" and week_day == "вторник" and day_type == "раб"






    













    




station_name == "ТУЧКОВО" and week_day == "четверг" and day_type == "раб"






    













    




station_name == "ТУЧКОВО" and week_day == "среда" and day_type == "раб"






    













    




station_name == "ТУЧКОВО" and week_day == "пятница" and day_type == "раб"






    













    




station_name == "ТУЧКОВО" and week_day == "понедельник" and day_type == "раб"






    













    




station_name == "БЕГОВАЯ" and week_day == "суббота" and day_type == "вых"






    













    




station_name == "БЕГОВАЯ" and week_day == "воскресенье" and day_type == "вых"






    













    




station_name == "БЕГОВАЯ" and week_day == "вторник" and day_type == "раб"






    













    




station_name == "БЕГОВАЯ" and week_day == "четверг" and day_type == "раб"






    













    




station_name == "БЕГОВАЯ" and week_day == "среда" and day_type == "раб"






    













    




station_name == "БЕГОВАЯ" and week_day == "пятница" and day_type == "раб"






    













    




station_name == "БЕГОВАЯ" and week_day == "понедельник" and day_type == "раб"






    













    




station_name == "БАКОВКА" and week_day == "суббота" and day_type == "вых"






    













    




station_name == "БАКОВКА" and week_day == "воскресенье" and day_type == "вых"






    













    




station_name == "БАКОВКА" and week_day == "вторник" and day_type == "раб"






    













    




station_name == "БАКОВКА" and week_day == "четверг" and day_type == "раб"






    













    




station_name == "БАКОВКА" and week_day == "среда" and day_type == "раб"






    













    




station_name == "БАКОВКА" and week_day == "пятница" and day_type == "раб"






    













    




station_name == "БАКОВКА" and week_day == "понедельник" and day_type == "раб"






    













    




station_name == "ГОЛИЦЫНО" and week_day == "суббота" and day_type == "вых"






    













    




station_name == "ГОЛИЦЫНО" and week_day == "воскресенье" and day_type == "вых"






    













    




station_name == "ГОЛИЦЫНО" and week_day == "вторник" and day_type == "раб"






    













    




station_name == "ГОЛИЦЫНО" and week_day == "четверг" and day_type == "раб"






    













    




station_name == "ГОЛИЦЫНО" and week_day == "среда" and day_type == "раб"






    













    




station_name == "ГОЛИЦЫНО" and week_day == "пятница" and day_type == "раб"






    













    




station_name == "ГОЛИЦЫНО" and week_day == "понедельник" and day_type == "раб"






    













    




station_name == "ЖАВОРОНКИ" and week_day == "суббота" and day_type == "вых"






    













    




station_name == "ЖАВОРОНКИ" and week_day == "воскресенье" and day_type == "вых"






    













    




station_name == "ЖАВОРОНКИ" and week_day == "вторник" and day_type == "раб"






    













    




station_name == "ЖАВОРОНКИ" and week_day == "четверг" and day_type == "раб"






    













    




station_name == "ЖАВОРОНКИ" and week_day == "среда" and day_type == "раб"






    













    




station_name == "ЖАВОРОНКИ" and week_day == "пятница" and day_type == "раб"






    













    




station_name == "ЖАВОРОНКИ" and week_day == "понедельник" and day_type == "раб"






    













    




station_name == "ОТРАДНОЕ СМОЛ." and week_day == "суббота" and day_type == "вых"






    













    




station_name == "ОТРАДНОЕ СМОЛ." and week_day == "воскресенье" and day_type == "вых"






    













    




station_name == "ОТРАДНОЕ СМОЛ." and week_day == "вторник" and day_type == "раб"






    













    




station_name == "ОТРАДНОЕ СМОЛ." and week_day == "четверг" and day_type == "раб"






    













    




station_name == "ОТРАДНОЕ СМОЛ." and week_day == "среда" and day_type == "раб"






    













    




station_name == "ОТРАДНОЕ СМОЛ." and week_day == "пятница" and day_type == "раб"






    













    




station_name == "ОТРАДНОЕ СМОЛ." and week_day == "понедельник" and day_type == "раб"






    













    




station_name == "ОДИНЦОВО" and week_day == "суббота" and day_type == "вых"






    













    




station_name == "ОДИНЦОВО" and week_day == "воскресенье" and day_type == "вых"






    













    




station_name == "ОДИНЦОВО" and week_day == "вторник" and day_type == "раб"






    













    




station_name == "ОДИНЦОВО" and week_day == "четверг" and day_type == "раб"






    













    




station_name == "ОДИНЦОВО" and week_day == "среда" and day_type == "раб"






    













    




station_name == "ОДИНЦОВО" and week_day == "пятница" and day_type == "раб"






    













    




station_name == "ОДИНЦОВО" and week_day == "понедельник" and day_type == "раб"






    













    




station_name == "МОСКВА БЕЛОРУССКАЯ" and week_day == "суббота" and day_type == "вых"






    













    




station_name == "МОСКВА БЕЛОРУССКАЯ" and week_day == "воскресенье" and day_type == "вых"






    













    




station_name == "МОСКВА БЕЛОРУССКАЯ" and week_day == "вторник" and day_type == "раб"






    













    




station_name == "МОСКВА БЕЛОРУССКАЯ" and week_day == "четверг" and day_type == "раб"






    













    




station_name == "МОСКВА БЕЛОРУССКАЯ" and week_day == "среда" and day_type == "раб"






    













    




station_name == "МОСКВА БЕЛОРУССКАЯ" and week_day == "пятница" and day_type == "раб"






    













    




station_name == "МОСКВА БЕЛОРУССКАЯ" and week_day == "понедельник" and day_type == "раб"






    

In [ ]:

    




    

In [ ]: