-
Notifications
You must be signed in to change notification settings - Fork 1
/
EvaluateCFRandom.py
76 lines (63 loc) · 2.67 KB
/
EvaluateCFRandom.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
import math, random, sys
from utils import parse
from techniques import Filter
"""
A wrapper around the random sampling evaluation technique which performs the
following:
- Randomly generates a `size` number of test cases such that the rating (userID,
itemID) != 99 (ie each test case represents prediction of a score actually
found in the matrix).
- For each generated test case, use the selected collborative filtering method
to predict the rating.
- Compare the predicted rating with the actual rating
- Output the results
- Compute the output of the Mean Absolute Error (MAE) achieved in your test
Inputs: method, size
- method: Collaborative filtering method to use
- size: number of test cases to generate
"""
def get_ratings(data, userID):
return data[userID]['ratings']
def get_val(data, userID, itemID):
return float(get_ratings(data, userID)[itemID])
def gen_tests(data, size):
tests = []
for i in xrange(size):
uid = random.randint(0, len(data)-1)
iid = random.randint(0, len(get_ratings(data, uid))-1)
while get_val(data, uid, iid) == 99:
iid = random.randint(0, len(get_ratings(data, uid))-1)
tests.append([uid,iid])
return tests
def print_evaluation(f, method, results):
print method
print " MAE: ",f.mean_absolute_error(results)
print " MSE: ",f.mean_squared_error(results)
print " RMSE: ",f.root_mean_squared_error(results)
print " NMAE: ",f.normalized_mean_absolute_error(results)
if __name__ == '__main__':
if len(sys.argv) != 3:
print 'Expected input format: python EvaluateCFList.py <method> <testList>'
else:
filename = 'data/jester-data-1.csv'
items = {}
users = {}
matrix = []
size = int(sys.argv[2])
matrix, users, items = parse(filename)
testData = gen_tests(users, size)
f = Filter(matrix, users, items)
method = sys.argv[1]
print "Starting predictions"
if method == 'all':
w_results = f.execute('weighted_sum', testData)
a_w_results = f.execute('adj_weighted_sum', testData)
c_w_results = f.execute('cosine_weighted_sum', testData)
c_a_w_results = f.execute('cosine_adj_weighted_sum', testData)
print_evaluation(f, "Weighted Sum", w_results)
print_evaluation(f, "Adjusted Weighted Sum", a_w_results)
print_evaluation(f, "Cosine Weighted Sum", c_w_results)
print_evaluation(f, "Cosine Adjusted Weighted Sum", c_a_w_results)
else:
results = f.execute(method, testData)
print_evaluation(f, method, results)