Page tree

This clinic outcomes page will be posted in a confluence space available to all Federal and contractor employees in the CCSQ Community

Original Story:

Author: Rob Christeson


Visualize Confluence/Syslog Anomaly Detection on QNOD dashboard
AC:
Validate data exists in influxdb
Create panels to visualize the predicted KPIs
Save and upload to s3 bucket for promoting to higher envs


Notes:

User is a system owner, or someone directly responsible for monitoring the system


5 Component metrics used in the prediction model. The modeling tool creates its own predictive model based on historical data, divided into two parts, one for historical analysis and one to compare its prediction algorithms agains to validate accuracy. The component metrics/KPIs are: Memory usage. active flow count, process count, cpu ut% disc free%


Grafana is end user interface

Revised Story(s):


As an owner of a system, I want to see a basic anomaly detection panel on the QNOD Dashboard so I know if the system I monitor is at risk of destabilizing


  • Verify that a simple anomaly warning is visible
  • Verify that the dashboard panel reads "No Anomalies" 
  • Verify that if an anomaly is detected, the warning reads "Anomaly Detected"


As an owner of a system, I want to see an Anomaly timeline panel on the QNOD Dashboard so I know I can track the a predicted anomaly and act to prevent it


  • Verify that a simple anomaly warning is visible
  • Verify that the dashboard panel reads "No Anomalies" 
  • Verify that if an anomaly is detected, the warning reads "Anomaly Detected"


As an owner of a system, I want to be alerted to a predicted anomaly so I don't have to watch the dashboard to know if a possible anomaly is predicted


  • Verify that an email(?) alert is generated when an anomaly is predicted
  • Verify that the the alert is generated when the anomaly prediction reaches a certainty threshold of 75%


Robert Christeson I took a stab at the our conversation about the work you described
regarding the Alerting Validation, not sure what the specific alerting, alert methods are, but you get the idea.


Original Story:

Author



Notes:


Revised Story(s):











A a medicare recipient user can log into our website so they can view claim information




  • No labels