def test_item_knn():
ibknn = pyreclab.ItemKnn(dataset='dataset/u1.base',
dlmchar=b'\t',
header=False,
usercol=0,
itemcol=1,
ratingcol=2)
ibknn.train(k=100, similarity='pearson')
pred = ibknn.predict('457', '443')
assert abs(pred - expected_prediction) < prediction_epsilon
ranking = ibknn.recommend('457', 5, includeRated=False)
assert ranking == expected_ranking
predlist, mae, rmse = ibknn.test(input_file='dataset/u1.test',
dlmchar=b'\t',
header=False,
usercol=0,
itemcol=1,
ratingcol=2,
output_file='predictions.csv')
assert abs(mae - expected_mae) < mae_epsilon
assert abs(rmse - expected_rmse) < rmse_epsilon
def main(k=100):
ibknn = pyreclab.ItemKnn(dataset='dataset/u1.base',
dlmchar=b'\t',
header=False,
usercol=0,
itemcol=1,
ratingcol=2)
print('-> training model')
start = time.clock()
ibknn.train(k=k, similarity='pearson')
end = time.clock()
print('training time: ' + str(end - start))
print('-> individual test')
pred = ibknn.predict('457', '443')
print('user 457, item 443, prediction ' + str(pred))
ranking = ibknn.recommend('457', 5, includeRated=False)
print('recommendation for user 457: ' + str(ranking))
print('-> prediction test')
start = time.clock()
predlist, mae, rmse = ibknn.test(input_file='dataset/u1.test',
dlmchar=b'\t',
header=False,
usercol=0,
itemcol=1,
ratingcol=2,
output_file='predictions.csv')
end = time.clock()
print('prediction time: ' + str(end - start))
print('MAE: ' + str(mae))
print('RMSE: ' + str(rmse))
print('-> recommendation test')
start = time.clock()
recommendList = ibknn.testrec(input_file='dataset/u1.test',
dlmchar=b'\t',
header=False,
usercol=0,
itemcol=1,
ratingcol=2,
topn=10,
output_file='ranking.json',
includeRated=False)
end = time.clock()
print('recommendation time: ' + str(end - start))
return mae, rmse
import time
import pyreclab
if __name__ == '__main__':
ibknn = pyreclab.ItemKnn(dataset='dataset/u1.base',
dlmchar=b'\t',
header=False,
usercol=0,
itemcol=1,
ratingcol=2)
print('-> training model')
start = time.clock()
ibknn.train(k=100, similarity='pearson')
end = time.clock()
print('training time: ' + str(end - start))
print('-> individual test')
pred = ibknn.predict('457', '443')
print('user 457, item 443, prediction ' + str(pred))
ranking = ibknn.recommend('457', 5, includeRated=False)
print('recommendation for user 457: ' + str(ranking))
print('-> prediction test')
start = time.clock()
predlist, mae, rmse = ibknn.test(input_file='dataset/u1.test',
dlmchar=b'\t',
header=False,
usercol=0,
prediction_filename = data_url + 'predictionsItemKnn' #5.csv'
ordenada_filename = data_url + 'ordenadasItemKnn' #5.csv'
print 'ItemKnn'
for i in [1, 5]: #range(1, data_chunks + 1):
f_t = training_filename + str(i) + ".txt"
f_p = probe_filename + str(i) + ".txt"
f_pred = prediction_filename + str(i) + ".csv"
f_ord = ordenada_filename + str(i) + ".csv"
print "Corriendo experimento ", i, "..."
print 'Entrenando...'
obj = pyreclab.ItemKnn(dataset=f_t,
dlmchar=b'\t',
header=True,
usercol=0,
itemcol=1,
ratingcol=2)
obj.train(10, 'pearson')
print 'Prediciendo...'
#prediction = obj.predict( "630685", "1")
#ranking = obj.recommend( "630685", 10, True)
#print prediction
#print ranking
predictionList, mae, rmse = obj.test(input_file=f_p,
dlmchar=b'\t',
header=False,
usercol=0,
itemcol=1,
ratingcol=2,