import pandas as pd

schedule = pd.DataFrame.from_csv("phillies.csv")

schedule.info()

schedule.head()

schedule.drop(["REMINDER_OFF", 
             "REMINDER_ON",
             "START_TIME_ET",
             "END_DATE",
             "END_DATE_ET",
             "END_TIME",
             "END_TIME_ET",
             "REMINDER_TIME",
             "REMINDER_TIME_ET",
             "SHOWTIMEAS_FREE",
             "SHOWTIMEAS_BUSY",
             "REMINDER_DATE"], axis=1, inplace=True)
schedule.head()

schedule.DESCRIPTION.head()

description = schedule.DESCRIPTION[2]
print description

def tv_stations_from_description(description):
    """Return a list of television stations embedded in the given description."""
    return [station.strip() for station in description.split(":")[1].split("-----")[0].split("--")]

result = tv_stations_from_description(description)
print result
assert(len(result) == 2)

description = schedule.DESCRIPTION[0]
print description
result = tv_stations_from_description(description)
print result
assert(len(result) == 1)

stations_series = schedule.DESCRIPTION.apply(
    lambda description: [station.strip() for station in 
                         description.split(":")[1].split("-----")[0].split("--")])
stations_series

set([station for stations in stations_series.values for station in stations])

schedule[(schedule.DESCRIPTION.str.contains("NBC 10")) | 
         (schedule.DESCRIPTION.str.contains("FOX"))]