Persisting merged DataFrame

If your data can fit into memory, either on a workstation or a cluster, persisting the Dask DataFrame with the .persist() method can be useful. In particular, persisting DataFrames can provide significant speed-up because the time-consuming work of reading from disk is performed only once.

The following function, which computes the percentage of flights delayed by weather, has been defined for you:

def percent_delayed(df):
    return (df['WEATHER_DELAY'].count() / len(df)) * 100

Your job is to time the execution of that function using the weather_delays DataFrame. You'll then call .persist() on the DataFrame and compare execution times of the function with the persisted DataFrame.

Note: This exercise will take several seconds to execute.

Print the time in milliseconds to compute the result of percent_delayed with weather_delays.
Call weather_delays.persist() and assign the result to persisted_weather_delays.
Print the time in milliseconds to compute the result of percent_delayed with persisted_weather_delays.

Take Hint (-30 XP)

Exercise

Persisting merged DataFrame

Instructions