def evaluate(output=True):
ratio = 0.1
k_value = 3
data_set, labels = get_data_set_from_file("dating.dataset")
data_set, m, l = knn.auto_normalize(data_set)
row_count = data_set.shape[0]
row_test_count = int(ratio * row_count)
test_data_set = data_set[0:row_test_count,:]
classify_fun = knn.gen_classify_fun(
[data_set[row_test_count:row_count,:],
labels[row_test_count:row_count]], k_value)
def is_right(row, label):
result = classify_fun(row)
if (result != label):
if output:
print "the classifier came back with: %d, the real answer is: %d" \
% (result, label)
return False
return True
result = [is_right(row.tolist()[0], label)
for label, row in zip(labels, test_data_set)]
error_count = result.count(False)
error_rate = error_count/float(row_test_count)
if output:
print "the total error rate is: %f (%d / %d)" \
% (error_rate, error_count, row_test_count)
return error_rate
def run_evaluate(output=True):
k_value = 3
classify_fun = knn.gen_classify_fun(
get_handwriting_dataset('training_digits'), k_value)
test_pathname = 'test_digits'
test_dataset, test_labels = get_handwriting_dataset(
test_pathname)
error_list, result = knn.evaluate(
classify_fun, test_dataset, test_labels, k_value)
if output:
print "error info:", error_list
pp([(os.listdir(test_pathname)[index], "actual:"+actual, "expected:"+expected)
for index, actual, expected, content in result])
return error_list, result
