Pivot pandas Dataframes with non-numeric values

Setup

# Import libraries
import pandas as pd
import numpy as np

# Sample DataFrame
df = pd.DataFrame({
    'advertiser':['Alice', 'Alice', 'Alice', 'Alice', 'Bob', 'Bob', 'Bob', 'Charlie', 'Charlie', 'Charlie'],
    'keyword': ['screen', 'phone', 'laptop', 'keyboard', 'phone', 'tablet', 'flat', 'screen', 'phone', 'computer'],
    'clicks': np.random.randint(low=20, high=50, size=10)
})
df

	advertiser	keyword	clicks
0	Alice	screen	34
1	Alice	phone	46
2	Alice	laptop	34
3	Alice	keyboard	36
4	Bob	phone	24
5	Bob	tablet	33
6	Bob	flat	25
7	Charlie	screen	35
8	Charlie	phone	25
9	Charlie	computer	35

Pivot keywords to columns

# Pivot keywords to columns
(
    df
    .pivot(index='advertiser', columns='keyword', values='clicks')
)

keyword	computer	flat	keyboard	laptop	phone	screen	tablet
advertiser
Alice	NaN	NaN	36.0	34.0	46.0	34.0	NaN
Bob	NaN	25.0	NaN	NaN	24.0	NaN	33.0
Charlie	35.0	NaN	NaN	NaN	25.0	35.0	NaN

# Pivot keywords to columns, ranked by clicks
(
    df
    .sort_values('clicks', ascending=False)
    .groupby('advertiser')
    .agg(list)
    .apply(lambda x: pd.Series(x['keyword']), axis=1)
)

	0	1	2	3
advertiser
Alice	phone	keyboard	screen	laptop
Bob	tablet	flat	phone	NaN
Charlie	screen	computer	phone	NaN