pandas - create sparse matrix for collaborative filtering

Question

I have a pandas dataframe like this:

user_id	music_id	rating
A	a	5
B	a	3

and I would like to create a sparse matrix from it, putting music_id as column and user_id as a row like this:

->

	a	b
A	5
B	3

what would be the way to achieve it by using pandas or numpy for this task?

Answer 1

If we suppose you have the following dataset as describe in your question:

d = {'user_id':['A','B'],'music_id':['a','a'],'rating':[5,3]}
df = pd.DataFrame(d)

Then you can do:

df.set_index(['user_id','music_id']).unstack(level=-1).rating

or, equivalently:

pd.pivot_table(df,values='rating',index='user_id',columns=['music_id'])