Sampling one record per unique value (pandas, python)

Based on number of rows per user this might be faster:

df.sample(frac=1).drop_duplicates(['User'])

df1_user_sample_one = df1.groupby('User').apply(lambda x:x.sample(1)) 

Using DataFrame.groupby.apply and lambda function to sample 1

This is what you want:

df1.groupby('User').apply(lambda df: df.sample(1))

Without the extra index:

df1.groupby('User', group_keys=False).apply(lambda df: df.sample(1))