def run(self, edit):
# Reload settings
settings.load()
current_selection = self.view.sel()
# Use the end of the current selection for the search, or use 0 if nothing is selected
begin = 0
if len(current_selection) > 0:
begin = self.view.sel()[-1].end()
conflict_region = find_conflict(self.view, begin)
if conflict_region is None:
return
# Add the region to the selection
self.view.show_at_center(conflict_region)
current_selection.clear()
current_selection.add(conflict_region)
def run(self):
# Reload settings
settings.load()
# Ensure git executable is available
if not self.git_executable_available():
sublime.error_message(msgs.get('git_executable_not_found'))
return
self.git_repo = self.determine_git_repo()
if not self.git_repo:
sublime.status_message(msgs.get('no_git_repo_found'))
return
conflict_files = self.get_conflict_files()
if not conflict_files:
sublime.status_message(msgs.get('no_conflict_files_found', self.git_repo))
return
self.show_quickpanel_selection(conflict_files)
def run(self, edit, keep):
# Reload settings
settings.load()
current_selection = self.view.sel()
# Use the begin of the current selection for the search, or use 0 if nothing is selected
begin = 0
if len(current_selection) > 0:
begin = current_selection[0].begin()
conflict_region = find_conflict(self.view, begin)
if conflict_region is None:
return
replace_text = extract(self.view, conflict_region, keep)
if not replace_text:
replace_text = ""
self.view.replace(edit, conflict_region, replace_text)
def plugin_loaded():
settings.load()
def main(argv):
#===== Initialization =====#
folder = "/ne_data/"
# Parse arguments
parser = argparse.ArgumentParser()
parser.add_argument('hostname',
type=str,
help="IP address or hostname of Z-way server host")
parser.add_argument('-u',
'--username',
type=str,
help="Username for Z-way server host")
parser.add_argument('-p',
'--password',
type=str,
help="Password for Z-way server host")
parser.add_argument('-s',
'--sound',
action='store_true',
help="use sound as a feature in analysis")
parser.add_argument('-f',
'--settings_file',
type=str,
help="load analysis settings from file")
#parser.add_argument('-b', '--backup', action='store_true', help="start training on backup data")
parser.add_argument(
'-t',
'--time_allign',
action='store_true',
help="collect data only at times which are multiples of the granularity"
)
parser.add_argument('-o',
'--collect_only',
action='store_true',
help="collect data but do not run analysis")
args = parser.parse_args(argv[1:])
# Initialize Zway server
host = args.hostname
if args.username and args.password:
zserver = zway.Server(host,
username=args.username,
password=args.password)
else:
zserver = zway.Server(host)
# Use default settings or read settings from settings file
if (args.settings_file == None):
settings_dict = {
"prefix": "ne_data",
"granularity": 60,
"training_window": 120,
"training_interval": 60,
"ema_alpha": 1.0,
"severity_omega": 1.0,
"severity_lambda": 3.719,
"auto_regression": 0.0
}
else:
try:
settings_dict = settings.load(args.settings_file)
except Exception as error:
print "Error reading settings file.", error
print " "
exit(1)
# Initialize Algo class
prefix = settings_dict['prefix']
granularity = int(settings_dict['granularity'])
training_window = int(settings_dict['training_window'])
training_interval = int(settings_dict['training_interval'])
ema_alpha = float(settings_dict['ema_alpha'])
severity_omega = float(settings_dict['severity_omega'])
severity_lambda = float(settings_dict['severity_lambda'])
auto_regression = int(settings_dict['auto_regression'])
feature_names = zserver.device_IDs()
num_features = len(feature_names)
print "Num features: ", num_features
print "w = %.3f, L = %.3f" % (severity_omega, severity_lambda)
print "alpha: %.3f" % ema_alpha
algo = Algo(num_features, training_window, training_interval)
algo.set_severity(severity_omega, severity_lambda)
algo.set_EWMA(ema_alpha)
# Two Datalogs: one for data and one for results
feature_names.append('total_power')
print(feature_names)
data_log = Datalog(folder, prefix, feature_names)
results_header = ['target', 'prediction', 'anomaly']
results_log = Datalog(folder, prefix + '_results', results_header)
# Timing procedure
granularity = settings_dict['granularity']
goal_time = int(time.time())
if args.time_allign:
goal_time += granularity - (int(time.time()) % granularity)
#===== Analysis =====#
while (True):
# Timing procedures
while goal_time > time.time():
time.sleep(0.2)
goal_time = goal_time + granularity
# Data collection
print "Recording sample at {}".format(goal_time)
features = get_features(zserver)
power = get_power()
features.append(power)
data_log.log(features[:], goal_time)
# Do not run analysis if only collecting data
if (args.collect_only): continue
features = np.array(features).flatten()
target, pred, anomaly, zscore = algo.run(features)
if (anomaly != None):
results_log.log([target, pred, float(anomaly)])
print target, pred, anomaly
else:
print target, pred
# Clean-up if necessary
print "Ending analysis"
def main(argv):
#np.seterr(all='print')
# Parse arguments
parser = argparse.ArgumentParser()
parser.add_argument('infile', type=str, help="name of input CSV file")
parser.add_argument('outfile', type=str, help="name of output CSV file")
parser.add_argument('-f',
'--settings_file',
type=str,
help="load analysis settings from file")
args = parser.parse_args(argv[1:])
infile = args.infile
outfile = args.outfile
print("Starting analysis on %s..." % infile)
print("Results will be recorded in %s..." % outfile)
# Use default settings or read settings from settings file
if (args.settings_file == None):
settings_dict = {
"granularity": 60,
"training_window": 1440,
"training_interval": 60,
"ema_alpha": 1.0,
"severity_omega": 1.0,
"severity_lambda": 3.719,
"auto_regression": 0.0
}
else:
try:
settings_dict = settings.load(args.settings_file)
except Exception as error:
print "Error reading settings file.", error
print " "
exit(1)
training_window = int(settings_dict['training_window'])
training_interval = int(settings_dict['training_interval'])
ema_alpha = float(settings_dict['ema_alpha'])
severity_omega = float(settings_dict['severity_omega'])
severity_lambda = float(settings_dict['severity_lambda'])
auto_regression = int(settings_dict['auto_regression'])
# Collect data from CSV file
df = pd.read_csv(infile)
df = df.set_index(df.columns[0])
timestamps = df.index.values
targets = df[df.columns[-1]]
features = df[df.columns[:-1]]
# Pre-processing
features, removed = filter_low_variance(features)
print "\nRemoving features due to low variance:"
for f in removed:
print f
num_features = len(features.columns) + auto_regression
print "\nThe following {} features will be used:".format(num_features)
for f in features.columns:
print f
if auto_regression > 0:
print "Auto-regression: {}\n".format(auto_regression)
# Add attacks to the data (power is last row)
# Used for F1 calculation
y = np.matrix(targets.values).T
start = len(y) / 2
y[start:], ground_truth = add_attacks(y[start:], 3, 30, 500)
ground_truth = set([(i + start) for i in ground_truth])
detected = set()
X = np.matrix(features.values)
X = scale_features(X)
X = add_auto_regression(X, y, auto_regression)
X = np.concatenate((X, y), 1)
print ""
# Initialize Algo class
algo = Algo(num_features, training_window, training_interval)
algo.set_severity(severity_omega, severity_lambda)
algo.set_EWMA(ema_alpha)
# Output lists
results = [['timestamp', 'target', 'prediction', 'anomaly', 'zscore']]
print "Num features: ", num_features
print "w = %.3f, L = %.3f" % (severity_omega, severity_lambda)
print "alpha: %.3f" % ema_alpha
#==================== ANALYSIS ====================#
print "Beginning analysis..."
count = 0
start_time = time.time()
for count in range(len(timestamps)):
# Get the next row of data
cur_time = timestamps[count]
if (count % 120 == 0):
print "Trying time %s" % \
dt.datetime.fromtimestamp(cur_time).strftime(DATE_FORMAT)
new_data = np.ravel(X[count, :])
new_data = np.around(new_data, decimals=2)
target, prediction, anomaly, zscore = algo.run(new_data) # Magic!
if (prediction != None):
results.append([cur_time, target, prediction, anomaly, zscore])
if anomaly:
detected.add(count)
count += 1
#==================== RESULTS ====================#
print "Runtime: %.4f" % (time.time() - start_time)
# Save data for later graphing
with open(outfile, 'wb') as csvfile:
writer = csv.writer(csvfile)
writer.writerows(results)
# Remove headers for analysis and graphing
timestamps, targets, predictions, anomalies, zscores = zip(*results)
timestamps = timestamps[1:]
targets = targets[1:]
predictions = predictions[1:]
anomalies = anomalies[1:]
zscores = zscores[1:]
precision, recall, f1_score = f1_scores(detected, ground_truth)
PMSE_smoothed, PMSE, Re_MSE, SMSE = error_scores(
targets, predictions) #Remove headers
print "{}: {:.4f}".format("RMSE-score (smoothed)", PMSE_smoothed)
print "{}: {:.4f}".format("RMSE-score (raw)", PMSE)
print "{}: {:.4f}".format("Relative MSE", Re_MSE)
print "{}: {:.4f}".format("SMSE", SMSE)
print "{}: {:.4f}".format("Precision", precision)
print "{}: {:.4f}".format("Recall", recall)
print "{}: {:.4f}".format("F1-Score", f1_score)
print "Ending analysis. See %s for results." % outfile
"""
plt.figure()
plt.subplot(211)
plt.plot(timestamps, targets, timestamps, predictions)
plt.subplot(212)
plt.plot(timestamps, zscores)
plt.title("Targets and Predictions")
plt.tight_layout()
plt.show()
"""
"""
plt.subplot(312)
error = [targets[i] - predictions[i] for i in range(len(targets))]
plt.hist(np.ravel(error), 250, facecolor='green', alpha=0.75)
plt.axis([-1000, 1000, 0, 10000])
plt.title("Distribution of Errors")
plt.subplot(313)
plt.hist(np.ravel(pvalues), 50, facecolor='green', alpha=0.75)
plt.title("Distribution of P-Values")
plt.tight_layout()
plt.show()
"""
return results
