Okay, so what you're trying to do (probably) is solved by running a join using the 'Datapoints' as the index on the same dataframe. It's probably not efficient, but you'd do something like:

df2 = df
df_joined = df.join(other=df2, on='Datapoints', how='Left').selectExpr('df.Location as `Location A`, 'df2.Location as `LocationB`')
df_joined.groupBy('Location A', 'Location B').agg(count('Location B').alias('Count'))

 I'm not sure this syntax works exactly as is, since I've never tried joining a table to itself.