Coding interview (StrataScratch)
Table of contents
Coding Questions
Easy
Python
Amazon - Return unique list of unique id of April or May
| transaction_id | signup_id | transaction_start_date | amt |
|---|---|---|---|
| int | int | datetime | float |
Question:
Solution:
transactions["signup_id"][transactions["transaction_start_date"].dt.month.isin([4,5])].unique()
'''
Final returned Panda Series has NO column name
- transaction_start_date contains YYYY-MM-dd -> we only care April and May, so use .isin([a,b]) -> returns series of True False in the second []
- signup_id: is the target final column to be returned
- unique(): returns only unique value + reset column name
'''
Spotify - Aggregate Listening Data (ID: 10367)
| user_id | song_id | listen_duration |
|---|---|---|
| int | int | float |
Question: Capture spotify user habits
- Per user, find the total listening time and the count of unqiue songs they listen to
- return total listening time as min not seconds
- return columns should be: user_id, total_listen_duration, unique_song_count
Solution:
- First set all missing listening duration to 0 (aka: replace NaN values -> 0)
- set all exisiting listening duration to nearest min
- groupby user_id and reset_index so it becomes the main column
# Replace NaN values -> 0:
listening_habits['listen_duration'].fillna(0,inplace=True)
# groupby user_id and set total_listen_duration from sec -> mins
df = listning_habits.groupby('user_id').agg(
total_listen_duration=('listen_duration', 'sum'),
unique_song_count=('song_id', 'nunique')
).reset_index()
df['total_listen_duration'] = df['total_listen_duration'].apply(lambda x: round(x/60)
df
'''
Final returned Panda Series has NO column name
- transaction_start_date contains YYYY-MM-dd -> we only care April and May, so use .isin([a,b]) -> returns series of True False in the second []
- signup_id: is the target final column to be returned
- unique(): returns only unique value + reset column name
'''