def optimize_machine(paths): # optimizes over all the timeseries provided in the path list
# returns a list of algorithms and a list of corresponding anomalies
anomaly_dict= dict()
anomaly_list= list()
index= 1
for path in paths:
anomaly_dict[path]= dict()
print "finding anomaly for path no." + str(index), path
index+= 1
for algo, feature, w_size in algo_iter():
# TODO: replace with my_dict
try:
var= anomaly_dict[path][algo]
except KeyError:
anomaly_dict[path][algo]= dict()
try:
var= anomaly_dict[path][algo][feature]
except KeyError:
anomaly_dict[path][algo][feature]= dict()
anomaly_list.append(gateway.get_anomalies(path, algo, feature, percent=2, mul_dev= 3, window_size= w_size))
print len(anomaly_list[-1])
anomaly_dict[path][algo][feature][w_size]= anomaly_list[-1]
final_anomalies, weights= aggregate(anomaly_list, ratio= 0.01)
min_algo= list()
anomalies= list()
for path in paths:
min_algo.append(find_optimal_algo(anomaly_dict[path], final_anomalies))
a= min_algo[-1]
anomalies.append(anomaly_dict[path][a[0]][a[1]][a[2]])
return min_algo, anomalies, final_anomalies, weights
def ts_majority_vote(path, ratio=0.005):
anomaly_list = list()
for algo, feature, w_size in algo_iter():
anomaly_list.append(
gateway.get_anomalies(path,
algo,
feature,
window_size=w_size,
percent=10,
mul_dev=1.5))
final_anomalies, weights = aggregate(anomaly_list, ratio)
return final_anomalies
def optimize_machine(
paths): # optimizes over all the timeseries provided in the path list
# returns a list of algorithms and a list of corresponding anomalies
anomaly_dict = dict()
anomaly_list = list()
index = 1
for path in paths:
anomaly_dict[path] = dict()
print "finding anomaly for path no." + str(index), path
index += 1
for algo, feature, w_size in algo_iter():
# TODO: replace with my_dict
try:
var = anomaly_dict[path][algo]
except KeyError:
anomaly_dict[path][algo] = dict()
try:
var = anomaly_dict[path][algo][feature]
except KeyError:
anomaly_dict[path][algo][feature] = dict()
anomaly_list.append(
gateway.get_anomalies(path,
algo,
feature,
percent=2,
mul_dev=3,
window_size=w_size))
print len(anomaly_list[-1])
anomaly_dict[path][algo][feature][w_size] = anomaly_list[-1]
final_anomalies, weights = aggregate(anomaly_list, ratio=0.01)
min_algo = list()
anomalies = list()
for path in paths:
min_algo.append(find_optimal_algo(anomaly_dict[path], final_anomalies))
a = min_algo[-1]
anomalies.append(anomaly_dict[path][a[0]][a[1]][a[2]])
return min_algo, anomalies, final_anomalies, weights
def ts_majority_vote(path, ratio= 0.005):
anomaly_list= list()
for algo, feature, w_size in algo_iter():
anomaly_list.append(gateway.get_anomalies(path, algo, feature, window_size= w_size, percent= 10, mul_dev= 1.5))
final_anomalies, weights= aggregate(anomaly_list, ratio)
return final_anomalies
