This lesson's interactive features are locked, please to keep using them
Event Counting
The most fundamental question in event analytics: how many times did the thing happen?
Counting sounds trivial, but it often comes down to how you slice and group the data. Given our "events" DataFrame and event_type column, we can count in several ways. By event type:
print(events["event_type"].value_counts())
Which gives you:
page_view 4521
login 1893
purchase 312
signup 287
Name: event_type, dtype: int64
The value_counts() method gives you the overall distribution. Are most events page views? That's normal. Are most events purchases? Something is probably wrong with your data... there's no way your boss's product is that popular...
Count Per User
You can use the groupby() method to count the total events by user:
print(events.groupby("user_id")["event_type"].count())
Which prints:
user_id
101 73
202 23
532 45
...
Name: event_type, dtype: int64
You can also add a filter to count specific event types per user:
print(events[events["event_type"] == "purchase"].groupby("user_id").size())
Which prints:
user_id
101 23
202 5
...
dtype: int64
Assignment
SnackStack's product team wants a quick snapshot of event activity: how many of each event type, plus which one happens most.
Complete the get_event_summary function. It accepts an events DataFrame with an event_type column and returns a dictionary mapping each event type to its count, with an extra "most_common" key.