def prepare_test_set(self, lower, upper):
m = utils.get_itemID_between_intervals(self.cf_matrix, \
lower=lower, upper=upper)
s = self.test_set.keys()
removes = set(s).difference(m)
for element in removes:
self.test_set.pop(element)
def test_TopN_intervals(self, N=100, interval=100, min_item=20):
mat = self.cf_matrix
m = mat.shape[0] # 3417x6000
low = up = 0
values = []
inters = []
while m >= up:
up = up + interval
movies_compare = utils.get_itemID_between_intervals(mat, low, up)
num_movies = len(movies_compare)
if num_movies >= min_item:
mean = sum(self.take_hits(N=N, movies=movies_compare)) / float(num_movies)
values.append(mean)
#print "%d - %d | Number of Movies: %d" % (low, up, num_movies)
#print mean
print "%d-%d %f" % (low+1, up, mean)
inters.append((low+up)/2)
low = up
TopNCorrExperiment.draw_topN_corr(inters, values, N=N)
def test_category_accuracy_interval(self, interval=100, N=100, min_item=20):
# If harsh is "True" then every genre(s) of a movie should be fitted
# to other movie.
cat_data, mid_dict = create_category_database()
try:
mat = self.cf_matrix
except AttributeError:
self.lightweight_setup()
mat = self.cf_matrix
m = mat.shape[0]
low = up = 0
values = []
inters = []
while m >= up:
up = up + interval
movies_compare = utils.get_itemID_between_intervals(mat, low, up)
num_movies = len(movies_compare)
if num_movies >= min_item:
[total, jac_list] = self.take_hits(cat_data, mid_dict, N=N, movies=movies_compare)
mean = total
values.append(mean)
#print "%d - %d | Number of Movies: %d" % (low, up, num_movies)
#print mean
print "%d-%d %f" % (low+1, up, mean)
inters.append((low+up)/2)
low = up
CategoryCorrExperiment.draw_catN_corr(inters, values, N=N)
return (inters, values)