recommend one and all users

recommendations.py

def recommend(user_id):
    '''recommend N items to user'''
    sparse_user_item = load_npz("sparse_user_item.npz")

    with open('model.sav', 'rb') as pickle_in:
        model = pickle.load(pickle_in)

    recommended, _ =  zip(*model.recommend(user_id, sparse_user_item))

    return recommended, map_movies(recommended)


def recommend_all_users():
    '''recommend N items to all users'''
    sparse_user_item = load_npz("sparse_user_item.npz")

    with open('model.sav', 'rb') as pickle_in:
        model = pickle.load(pickle_in)

    # numpy array with N recommendations for each user
    # remove first array, because those are the columns
    all_recommended = model.recommend_all(user_items=sparse_user_item, N=10, 
        recalculate_user=False, filter_already_liked_items=True)[1:]

    # create a new Pandas Dataframe with user_id, 10 recommendations, for all users
    df = pd.read_csv('ml-1m/users.dat', delimiter='::', header=None, 
        names=['user_id', 'gender', 'agerange', 'occupation', 'timestamp'], engine='python')
    df = df.drop(['gender', 'agerange', 'occupation', 'timestamp'], axis=1)
    df[['rec1', 'rec2', 'rec3', 'rec4', 'rec5', 'rec6', 'rec7', 'rec8', 'rec9', 'rec10']] = pd.DataFrame(all_recommended)
    df.to_pickle("all_recommended.pkl")

    '''melt dataframe into SQL format for Django model
    melted = df.melt(id_vars=['user_id'], var_name='order', value_name='recommendations',
        value_vars=['rec1', 'rec2', 'rec3', 'rec4', 'rec5', 'rec6', 'rec7', 'rec8', 'rec9', 'rec10'])
    melted['order'] = melted.order.str[3:]
    print(melted.sort_values(by=['user_id', 'order']))
    melted.to_pickle('all_recommended_melted.pkl')
    '''

    return df