Experimenrs on Foursquare

import pandas as pd
import numpy as np
import codecs
import math

doc1 = codecs.open('dataset_TIST2015/dataset_TIST2015_Checkins.txt','rU','latin-1')
checkins = pd.read_csv(doc1, delimiter='	')
checkins.columns=['userId', 'venueId', 'timeUTC', 'timeOffset']

len(checkins['userId'].unique())
len(checkins['venueId'].unique())

from sklearn.preprocessing import LabelEncoder
from scipy.sparse import csr_matrix

venueIdencoder = LabelEncoder().fit(checkins['venueId'])
userIdencoder = LabelEncoder().fit(checkins['userId'])

checkins['venueIdencoded'] = venueIdencoder.transform(checkins['venueId'])
n_venues = len(venueIdencoder.classes_)

from sklearn.cross_validation import train_test_split

train_df, test_df = train_test_split(checkins, train_size = 0.8)

train = csr_matrix((np.ones(train_df.shape[0]), (train_df.userId, train_df.venueIdencoded)), shape=((train_df.userId.max()+1),n_venues))

test = csr_matrix((np.ones(test_df.shape[0]), (test_df.userId, test_df.venueIdencoded)), shape=((test_df.userId.max()+1),n_venues))

#print(test.nnz)
#print(train.nnz)

#print(test.max())
#print(train.max())

from lightfm import LightFM
from lightfm.evaluation import auc_score

NUM_THREADS = 1
NUM_COMPONENTS = 30
NUM_EPOCHS = 1
ITEM_ALPHA = math.exp(-6)

model = LightFM(loss='warp',
                item_alpha=ITEM_ALPHA,
                no_components=NUM_COMPONENTS)

model.fit(train,epochs=NUM_EPOCHS,num_threads=NUM_THREADS)


train_auc = auc_score(model, train,num_threads=NUM_THREADS).mean()
test_auc = auc_score(model, test,train_interactions=train,num_threads=NUM_THREADS).mean()

print("Train_auc is %f" %train_auc)
print("Test_aus is %f" %test_auc)

Some problems :

Expect to get a binary marix but no...

Here is the code in console:

train
Out[6]: 
<266910x3680125 sparse matrix of type '<class 'numpy.float64'>'
    with 12774460 stored elements in Compressed Sparse Row format>
train.data.max()
Out[7]: 520.0
train.data.min()
Out[8]: 1.0
test.data.max()
Out[9]: 140.0
test.data.mean()
Out[10]: 1.533210711390105
test.data.min()
Out[11]: 1.0        

 

and Running on cluster for one night but got no results showed...