def last_detect_sample(endpoint, key, request):
print('Sample of detecting whether the latest point in series is anomaly.')
client = AnomalyDetectorClient(endpoint, CognitiveServicesCredentials(key))
response = client.last_detect(request)
if response.is_anomaly:
print('The latest point is detected as anomaly.')
else:
print('The latest point is not detected as anomaly.')
def entire_detect_sample(endpoint, key, request):
print('Sample of detecting anomalies in the entire series.')
client = AnomalyDetectorClient(endpoint, CognitiveServicesCredentials(key))
response = client.entire_detect(request)
if True in response.is_anomaly:
print('Anomaly was detected from the series at index:')
for i in range(len(series)):
if response.is_anomaly[i]:
print(i)
else:
print('There is no anomaly detected from the series.')
def entire_dectect_sample(endpoint, key):
print('Sample of detecting anomalies in the entire series.')
client = AnomalyDetectorClient(endpoint, CognitiveServicesCredentials(key))
series = [
Point(timestamp=datetime(1962, 1, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 2, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 3, 1, tzinfo=timezone.utc), value=0),
Point(timestamp=datetime(1962, 4, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 5, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 6, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 7, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 8, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 9, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 10, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 11, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 12, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 1, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 2, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 3, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 4, 1, tzinfo=timezone.utc), value=0),
Point(timestamp=datetime(1963, 5, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 6, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 7, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 8, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 9, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 10, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 11, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 12, 1, tzinfo=timezone.utc), value=1),
]
data = Request(series=series,
granularity=Granularity.monthly,
sensitivity=95,
max_anomaly_ratio=0.25)
response = client.entire_detect(data)
if True in response.is_anomaly:
print('Anomaly was detected from the series at index:')
for i in range(len(series)):
if response.is_anomaly[i]:
print(i)
else:
print('There is no anomaly detected from the series.')
def last_detect(subscription_key):
print("Sample of detecting whether the latest point in series is anomaly.")
# Add your Azure Anomaly Detector subscription key to your environment variables.
endpoint = os.environ.get["ANOMALY_DETECTOR_ENDPOINT"]
try:
client = AnomalyDetectorClient(
endpoint, CognitiveServicesCredentials(subscription_key))
request = get_request()
response = client.last_detect(request)
if response.is_anomaly:
print("The latest point is detected as anomaly.")
else:
print("The latest point is not detected as anomaly.")
except Exception as e:
if isinstance(e, APIErrorException):
print("Error code: {}".format(e.error.code))
print("Error message: {}".format(e.error.message))
else:
print(e)
def last_detect(subscription_key):
print("Sample of detecting whether the latest point in series is anomaly.")
endpoint = "https://{}.api.cognitive.microsoft.com".format(
ANOMALYDETECTOR_LOCATION)
try:
client = AnomalyDetectorClient(
endpoint, CognitiveServicesCredentials(subscription_key))
request = get_request()
response = client.last_detect(request)
if response.is_anomaly:
print("The latest point is detected as anomaly.")
else:
print("The latest point is not detected as anomaly.")
except Exception as e:
if isinstance(e, APIErrorException):
print("Error code: {}".format(e.error.code))
print("Error message: {}".format(e.error.message))
else:
print(e)
def last_detect_sample(endpoint, key):
print('Sample of detecting whether the latest point in series is anomaly.')
client = AnomalyDetectorClient(endpoint, CognitiveServicesCredentials(key))
series = [
Point(timestamp=datetime(1962, 1, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 2, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 3, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 4, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 5, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 6, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 7, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 8, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 9, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 10, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 11, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1962, 12, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 1, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 2, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 3, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 4, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 5, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 6, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 7, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 8, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 9, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 10, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 11, 1, tzinfo=timezone.utc), value=1),
Point(timestamp=datetime(1963, 12, 1, tzinfo=timezone.utc), value=0),
]
data = Request(series=series,
granularity=Granularity.monthly,
sensitivity=95,
max_anomaly_ratio=0.25)
response = client.last_detect(data)
if response.is_anomaly:
print('The latest point is detected as anomaly.')
else:
print('The latest point is not detected as anomaly.')
def entire_detect(subscription_key):
print("Sample of detecting anomalies in the entire series.")
# Add your Azure Anomaly Detector subscription key to your environment variables.
endpoint = os.environ.get["ANOMALY_DETECTOR_ENDPOINT"]
try:
client = AnomalyDetectorClient(
endpoint, CognitiveServicesCredentials(subscription_key))
request = get_request()
response = client.entire_detect(request)
if True in response.is_anomaly:
print("Anomaly was detected from the series at index:")
for i in range(len(request.series)):
if response.is_anomaly[i]:
print(i)
else:
print("There is no anomaly detected from the series.")
except Exception as e:
if isinstance(e, APIErrorException):
print("Error code: {}".format(e.error.code))
print("Error message: {}".format(e.error.message))
else:
print(e)
def entire_detect(subscription_key):
print("Sample of detecting anomalies in the entire series.")
endpoint = "https://{}.api.cognitive.microsoft.com".format(
ANOMALYDETECTOR_LOCATION)
try:
client = AnomalyDetectorClient(
endpoint, CognitiveServicesCredentials(subscription_key))
request = get_request()
response = client.entire_detect(request)
if True in response.is_anomaly:
print("Anomaly was detected from the series at index:")
for i in range(len(request.series)):
if response.is_anomaly[i]:
print(i)
else:
print("There is no anomaly detected from the series.")
except Exception as e:
if isinstance(e, APIErrorException):
print("Error code: {}".format(e.error.code))
print("Error message: {}".format(e.error.message))
else:
print(e)
import os
# </imports>
# <initVars>
# Set the COGSRV_ACCOUNT_KEY environment variable on your local machine using the
# appropriate method for your preferred shell (Bash, PowerShell, Command Prompt, etc.)
SUBSCRIPTION_KEY = os.environ["COGSRV_ACCOUNT_KEY"]
ANOMALY_DETECTION_LOCATION = "westus2"
TIME_SERIES_DATA_PATH = ""
# </initVars>
# Create an Anomaly Detector client and add the
# <client>
base_url = "https://{}.api.cognitive.microsoft.com".format(ANOMALY_DETECTION_LOCATION)
client = AnomalyDetectorClient(base_url, CognitiveServicesCredentials(SUBSCRIPTION_KEY))
# </client>
# Load in the time series data file
# <loadDataFile>
series = []
data_file = pd.read_csv(TIME_SERIES_DATA_PATH, header=None, encoding='utf-8', parse_dates=[0])
for index, row in data_file.iterrows():
series.append(Point(timestamp=row[0], value=row[1]))
# </loadDataFile>
# Create a request from the data file
# <request>
request = Request(series=series, granularity=Granularity.daily)
import pandas as pd
import os
# </imports>
# <initVars>
# This sample assumes you have created an environment variable for your key and endpoint
SUBSCRIPTION_KEY = os.environ["ANOMALY_DETECTOR_KEY"]
ANOMALY_DETECTOR_ENDPOINT = os.environ["ANOMALY_DETECTOR_ENDPOINT"]
TIME_SERIES_DATA_PATH = "request-data.csv"
# </initVars>
# Create an Anomaly Detector client and add the
# <client>
client = AnomalyDetectorClient(ANOMALY_DETECTOR_ENDPOINT, CognitiveServicesCredentials(SUBSCRIPTION_KEY))
# </client>
# Load in the time series data file
# <loadDataFile>
series = []
data_file = pd.read_csv(TIME_SERIES_DATA_PATH, header=None, encoding='utf-8', parse_dates=[0])
for index, row in data_file.iterrows():
series.append(Point(timestamp=row[0], value=row[1]))
# </loadDataFile>
# Create a request from the data file
# <request>
request = Request(series=series, granularity=Granularity.daily)
def get_client_lazy(self):
from azure.cognitiveservices.anomalydetector import AnomalyDetectorClient
from msrest.authentication import CognitiveServicesCredentials
return AnomalyDetectorClient(
self.url, CognitiveServicesCredentials(self.get_secret()))
def detect_anomalies(X,
index,
interval,
overlap_size,
subscription_key,
endpoint,
granularity,
custom_interval=None,
period=None,
max_anomaly_ratio=None,
sensitivity=None,
timezone="UTC"):
"""Microsoft's Azure Anomaly Detection tool.
Args:
X (list):
Array containing the input value sequences.
index (list):
Array containing the input index sequences.
interval (int):
Integer denoting time span frequency of the data.
overlap_size (int):
Length of the values from previous sequence that overlaps with current sequnce.
subscription_key (str):
Resource key for authenticating your requests.
endpoint (str):
Resource endpoint for sending API requests.
granularity (str or Granularity):
Can only be one of yearly, monthly, weekly, daily, hourly or minutely.
Granularity is used for verify whether input series is valid. Possible values
include: 'yearly', 'monthly', 'weekly', 'daily', 'hourly', 'minutely'.
custom_interval (int):
Integer used to set non-standard time interval, for example, if the series
is 5 minutes, request can be set as `{"granularity":"minutely", "custom_interval":5}`.
If not given, `None` is used.
period (int):
Periodic value of a time series. If not given, `None` is used, and the API
will determine the period automatically.
max_anomaly_ratio (float):
Advanced model parameter, max anomaly ratio in a time series. If not given,
`None` is used.
sensitivity (int):
Advanced model parameter, between 0-99, the lower the value is, the larger
the margin value will be which means less anomalies will be accepted. If not given,
`None` is used.
timezone (str):
String indicating the timezone of the timestamps. If not given, will use UTC as
default. The format of the string should be complaint with ``pytz``
which can be found in http://pytz.sourceforge.net/.
Returns:
list:
Array containing start-index, end-index, score for each anomalous sequence.
Note that the API does not have an anomaly score, and so score is set to `None`.
"""
client = AnomalyDetectorClient(
endpoint, CognitiveServicesCredentials(subscription_key))
tz = pytz.timezone(timezone)
overlap = 0
result = list()
for x, idx in zip(X, index):
series = []
for i in range(len(x)):
idx_ = _convert_date(idx[i], tz)
series.append(Point(timestamp=idx_, value=x[i]))
request = Request(series=series,
granularity=granularity,
custom_interval=custom_interval,
period=period,
max_anomaly_ratio=max_anomaly_ratio,
sensitivity=sensitivity)
response = client.entire_detect(request)
if response.is_anomaly:
anomalous = response.is_anomaly[overlap:]
index_ = idx[overlap:]
result.extend(index_[anomalous])
overlap = overlap_size
return _convert_anomalies_to_contextual(result, interval)
