why my program is so slow?

[onestar S]

my training data :

total:750156

user-movie-ratings

i used item-based recommender

my code:

from scikits.crab.metrics import pearson_correlation, cosine_distances, euclidean_distances

from scikits.crab.similarities import ItemSimilarity 

from scikits.crab.models import MatrixPreferenceDataModel 

 

model = MatrixPreferenceDataModel(data_movies) 

item_sim_pearson = ItemSimilarity(model, pearson_correlation) 

from scikits.crab.recommenders.knn import ItemBasedRecommender 

item_recommender = ItemBasedRecommender(model, item_sim_pearson, with_preference=True)

i want to predict the (user , movie ) rating,

so  my test data is like this:

ID,user,movie

895537,5412,2683

total:250000

when i predict the rating, the program is so slow,

def predict(x):

    return item_recommender.estimate_preference(x.user, x.movie)

test_data['prediction'] = test_data.apply(predict, axis=1)

test_data is a dataframe of my test data...

i wonder that why it is slow, am i wrong?

please help me ...