notebook.community

Edit and run 





In [10]:



    


import pandas as pd
import os
import seaborn as sns
%matplotlib inline

Pedestrian counts

Get the data

http://bit.ly/1vm5nJ5

  • pedestrian_count_bridge.csv
  • pedestrian_count_station.csv

Things to do

  • Load the data and make sure it has a datetime index
  • Consolidate data into one dataframe
  • Plot a couple location counts together
  • Plot daily totals for pedestrians
  • Plot Monthly totals in individual sub-plots
  • Q) Which location has had the most traffic this year?
  • Q) Which location had the least traffic June?
  • Q) Which location had the most overall traffic at 8am on a weekday?
  • Plot a Histogram of total pedestrians per by hour on 24-7-2013 in Flagstaff Station
  • Q) Which location has had the biggest percentage change year on yearfrom 2012->2013
  • Q) What is the overall busiest day of the week for foot traffic
  • Q) What was the single busiest day, when and where
  • Q) Make a boxplot of the daily counts at each location

Load the data and make sure it has a datetime index





In [69]:



    


bridge = pd.read_csv('pedestrian_count_bridge.csv')
bridge = bridge.set_index(pd.DatetimeIndex(bridge.pop('datetime')))
station = pd.read_csv('pedestrian_count_station.csv')
station = station.set_index(pd.DatetimeIndex(station.pop('datetime')))

or





In [70]:



    


bridge = pd.read_csv('pedestrian_count_bridge.csv', index_col=0, parse_dates=[0])
station = pd.read_csv('pedestrian_count_station.csv', index_col=0, parse_dates=[0])



    











In [71]:



    


station.head()



    


















    
Out[71]:










  
    

      
      Flagstaff Station
      Southern Cross Station
      Flinders St Station Underpass
    

    

      datetime
      
      
      
    

  
  
    

      2012-01-01 00:00:00
       157
       325
       3726
    

    

      2012-01-01 01:00:00
       105
       246
       3082
    

    

      2012-01-01 02:00:00
        39
       131
       2784
    

    

      2012-01-01 03:00:00
        34
       135
       1915
    

    

      2012-01-01 04:00:00
        26
        56
        959
    

  




















In [21]:



    


bridge.head()



    


















    
Out[21]:










  
    

      
      Webb Bridge
      Princes Bridge
      Sandridge Bridge
    

    

      datetime
      
      
      
    

  
  
    

      2012-01-01 00:00:00
       325
       4171
        479
    

    

      2012-01-01 01:00:00
       169
       2843
       1011
    

    

      2012-01-01 02:00:00
        52
       2153
       1346
    

    

      2012-01-01 03:00:00
        58
        967
       1034
    

    

      2012-01-01 04:00:00
        18
        404
        532

Consolidate data into one dataframe





In [27]:



    


all_dat = bridge.join(station, how='inner')



    











In [72]:



    


all_dat.head()



    


















    
Out[72]:










  
    

      
      Webb Bridge
      Princes Bridge
      Sandridge Bridge
      Flagstaff Station
      Southern Cross Station
      Flinders St Station Underpass
    

    

      datetime
      
      
      
      
      
      
    

  
  
    

      2012-01-01 00:00:00
       325
       4171
        479
       157
       325
       3726
    

    

      2012-01-01 01:00:00
       169
       2843
       1011
       105
       246
       3082
    

    

      2012-01-01 02:00:00
        52
       2153
       1346
        39
       131
       2784
    

    

      2012-01-01 03:00:00
        58
        967
       1034
        34
       135
       1915
    

    

      2012-01-01 04:00:00
        18
        404
        532
        26
        56
        959

Plot a couple location counts together





In [73]:



    


bridge.plot()



    


















    
Out[73]:








<matplotlib.axes._subplots.AxesSubplot at 0x11798de50>

Plot daily totals for pedestrians





In [74]:



    


bridge.resample('D', how='sum').plot()



    


















    
Out[74]:








<matplotlib.axes._subplots.AxesSubplot at 0x1088c6110>

Plot Monthly totals in individual sub-plots





In [75]:



    


bridge.resample('M', how='sum').plot(subplots=True, figsize=(12, 10))



    


















    
Out[75]:








array([<matplotlib.axes._subplots.AxesSubplot object at 0x11436eed0>,
       <matplotlib.axes._subplots.AxesSubplot object at 0x116c4fa50>,
       <matplotlib.axes._subplots.AxesSubplot object at 0x1173b4650>], dtype=object)

Q) Which one has had the most traffic this year?

A) Princes Bridge





In [76]:



    


all_dat['2014':].sum().sort(inplace=False)



    


















    
Out[76]:








Webb Bridge                       20148205
Sandridge Bridge                  47711031
Southern Cross Station            64873641
Flagstaff Station                 96613925
Flinders St Station Underpass    106842183
Princes Bridge                   123709670
dtype: float64

Q) Which location had the least traffic June?

A) Webb Bridge





In [77]:



    


all_dat['2014-06'].sum().sort(inplace=False)
# OR all_dat[all_dat.index.month == 6]



    


















    
Out[77]:








Webb Bridge                       2150769
Sandridge Bridge                  5155049
Southern Cross Station            7294983
Flagstaff Station                10243709
Flinders St Station Underpass    11657484
Princes Bridge                   13526645
dtype: float64

Q) Which location had the most overall traffic at 8am on a weekday?

A) Flagstaff Station





In [78]:



    


all_dat[(all_dat.index.weekday < 5)
        & (all_dat.index.hour == 8)].sum().sort(inplace=False)



    


















    
Out[78]:








Webb Bridge                       263917
Sandridge Bridge                  501395
Princes Bridge                    792703
Southern Cross Station           1360759
Flinders St Station Underpass    1381658
Flagstaff Station                2518030
dtype: float64

Plot a Histogram of total pedestrians per by hour for July 2013 in Flagstaff Station





In [79]:



    


dt = station.ix['2014-07']['Flagstaff Station']
dt.groupby(dt.index.hour).sum().plot(kind='bar')



    


















    
Out[79]:








<matplotlib.axes._subplots.AxesSubplot at 0x118783990>

Q) Which location has had the biggest percentage change year on yearfrom 2012->2013

A) Flinders St





In [93]:



    


changes = (all_dat.resample('A', how='sum').diff()['2013'].values
           / all_dat.resample('A', how='sum')['2012'].values) * 100
pd.DataFrame(changes, columns=all_dat.columns)



    


















    
Out[93]:










  
    

      
      Webb Bridge
      Princes Bridge
      Sandridge Bridge
      Flagstaff Station
      Southern Cross Station
      Flinders St Station Underpass
    

  
  
    

      0
       13.952665
       11.12876
       6.574006
       0.877868
       14.384528
      -22.142738

Q) What is the overall busiest day of the week for foot traffic

A) Friday!





In [100]:



    


all_dat.sum(axis=1).groupby(all_dat.index.dayofweek).sum().sort(inplace=False, ascending=False)



    


















    
Out[100]:








6    55048427
5    55299765
2    77188589
0    79280516
3    82554956
1    85241115
4    88645828
dtype: float64

Q) What was the single busiest day, when and where

A) 2014-04-25 Princes Bridge





In [129]:



    


day_dat = all_dat.groupby(all_dat.index.date).sum()



    











In [133]:



    


for c in day_dat.columns:
    dt = day_dat[c].idxmax()
    mx = day_dat[c].max()
    print '{}\t{:>35}\t{:9.2f}'.format(dt, c, mx)



    


















    






2014-04-15	                        Webb Bridge	137280.00
2014-04-25	                     Princes Bridge	1253640.00
2014-09-26	                   Sandridge Bridge	311760.00
2014-10-04	                  Flagstaff Station	770448.00
2014-05-06	             Southern Cross Station	454944.00
2014-09-26	      Flinders St Station Underpass	723288.00

Make a boxplot of the daily counts at each location





In [158]:



    


import pylab as plt
import numpy as np
fig, ax = plt.subplots(1, figsize=(12, 10))
_ = day_dat.boxplot(grid=False, ax=ax)
ax.set_ylim(0, np.percentile(day_dat.values.flatten(), 82.5))
ax.set_xticklabels(day_dat.columns, rotation=90)



    


















    
Out[158]:








[<matplotlib.text.Text at 0x11bc1f950>,
 <matplotlib.text.Text at 0x1198134d0>,
 <matplotlib.text.Text at 0x11bf47410>,
 <matplotlib.text.Text at 0x11bf47b90>,
 <matplotlib.text.Text at 0x11bc1fad0>,
 <matplotlib.text.Text at 0x10fd137d0>]










    
























In [146]:



    


np.percentile(day_dat.values.flatten(), 90)



    


















    
Out[146]:








413932.80000000005

















In [ ]:

Content source: chausler/talks

Similar notebooks:

notebook.community | gallery | about