Pandas Workshop

cta.py

#Launch Python:
#cd anaconda
#open Launcher.app

ridership = pd.read_csv('https://raw.githubusercontent.com/zafia/talks/gh-pages/pandas-class/cta_ridership.csv')

ridership.loc[ridership.stationname.unique()]
ridership.loc[ridership.stationname.str.contains("Mo”)]
ridership.loc[ridership.stationname == "Morse"]
stations = ridership.groupby("stationname") => #<pandas.core.groupby.DataFrameGroupBy object at 0x102a3e310>
type(stations) => #pandas.core.groupby.DataFrameGroupBy
stations.rides.sum().max()
daily = ridership.groupby("date")
daily.rides.sum().idxmax() # most rides; lollapooloza
daily.rides.sum()

import matplotlib.pyplot as plt
%matplotlib inline # prints graphs for you
#a column in pandas is called a series


daily_ridership = pd.Series(daily.rides.sum())

daily_ridership.plot() #will plot a graph for you!!!
Out[42]: <matplotlib.axes._subplots.AxesSubplot at 0x10d1134d0>

ridership.daytype = ridership.daytype.map({"A" : "saturday", "U" : "sunday", "W" : "weekday"})
weekdays = ridership.loc[ridership.daytype == "weekday"]
weekdays.head(10)
Out[60]: 
     station_id         stationname                 date  daytype  rides
423       41670        Conservatory  2010-01-04 00:00:00  weekday    718
424       41660          Lake/State  2010-01-04 00:00:00  weekday  14813
425       41500      Montrose-Brown  2010-01-04 00:00:00  weekday   1963
426       41490            Harrison  2010-01-04 00:00:00  weekday   2287
427       41480       Western-Brown  2010-01-04 00:00:00  weekday   3398
428       41460   Irving Park-Brown  2010-01-04 00:00:00  weekday   2198
429       41450       Chicago/State  2010-01-04 00:00:00  weekday  11041
430       41440       Addison-Brown  2010-01-04 00:00:00  weekday   2090
431       41430                87th  2010-01-04 00:00:00  weekday   4558
432       41420  Addison-North Main  2010-01-04 00:00:00  weekday   4835



lolla_cta.py

ridership.date = pd.to_datetime(ridership.date)

lolla = ridership.loc[(ridership.date.dt.year == 2011) & (ridership.date.dt.month == 8)]

lolla.head(5)
Out[65]: 
       station_id              stationname       date  daytype  rides
81806       40730         Washington/Wells 2011-08-01  weekday   7420
81807       40720  East 63rd-Cottage Grove 2011-08-01  weekday   1443
81808       40710         Chicago/Franklin 2011-08-01  weekday   6591
81809       40700                  Laramie 2011-08-01  weekday   1390
81810       40700                  Laramie 2011-08-01  weekday   1381



lolla_grouped = lolla.groupby("date")

lolla_ridership = pd.Series(lolla_grouped.rides.sum())

lolla_ridership.plot()
Out[70]: <matplotlib.axes._subplots.AxesSubplot at 0x10d0cb290>
#shows awesome graph

stations_with_most_riders.py

station_ridership = pd.Series(stations.rides.sum().sort(inplace = False, ascending = False).head(5))

station_ridership
Out[72]: 
stationname
Lake/State            29458946
Clark/Lake            27744163
Chicago/State         26062944
Belmont-North Main    21712070
Fullerton             21485675
Name: rides, dtype: int64

station_ridership.plot(kind="bar") # shows awesome bar graph

s = station_ridership.loc[station_ridership != 0]