def experiment2(filename, ret_val):
mmre = []
for _ in xrange(30):
method = guo_random(filename)
training_data, testing_data = method.generate_test_data(ret_val)
assert(len(training_data[0]) == len(training_data[-1])), "Something is wrong"
assert(len(testing_data[0]) == len(testing_data[-1])), "Something is wrong"
mmre.append(generate_model(training_data, testing_data))
print round(np.mean(mmre) * 100, 3), round(np.std(mmre) * 100, 3),
def experiment1(filename, normalize=None):
if normalize is not None:
filename = normalize(filename)
mmre = []
for _ in xrange(30):
print "# ",
sys.stdout.flush()
method = what(filename)
training_data, testing_data = method.generate_test_data()
assert (len(training_data[0]) == len(
training_data[-1])), "Something is wrong"
assert (len(testing_data[0]) == len(
testing_data[-1])), "Something is wrong"
mmre.append(generate_model(training_data, testing_data))
print round(np.mean(mmre) * 100, 3), round(np.std(mmre) * 100, 3),
return len(training_data[0])
def experiment1(filename, normalize=None, feature_weights=False):
if normalize is not None:
name = filename.split('/')[-1]
norm_filename = "./NData/zscore/norm_" + name
mmre = []
for _ in xrange(30):
if feature_weights is True:
filename = add_feature_weights(norm_filename)
method = what(filename)
training_data, testing_data = method.generate_test_data()
assert(len(training_data[0]) == len(training_data[-1])), "Something is wrong"
assert(len(testing_data[0]) == len(testing_data[-1])), "Something is wrong"
mmre.append(generate_model(training_data, testing_data))
print round(np.mean(mmre) * 100, 3), round(np.std(mmre) * 100, 3),
return len(training_data[0])
test_independent.append(map(float, content[ti][:-1]))
test_dependent.append(float(content[ti][-1]))
assert (
len(test_independent) == len(test_dependent)), "something wrong"
train_indexes = range(len(train_independent))
shuffle(train_indexes)
selected_points_indep = [
train_independent[i] for i in xrange(len(train_indexes[:number]))
]
selected_points_dep = [
train_dependent[i] for i in xrange(len(train_indexes[:number]))
]
return [selected_points_indep,
selected_points_dep], [test_independent, test_dependent]
if __name__ == "__main__":
mmre = []
for _ in xrange(30):
method = guo_random("./Data/Apache_AllMeasurements.csv")
training_data, testing_data = method.generate_test_data()
assert (len(training_data[0]) == len(
training_data[-1])), "Something is wrong"
assert (len(testing_data[0]) == len(
testing_data[-1])), "Something is wrong"
mmre.append(generate_model(training_data, testing_data))
print np.median(mmre)