Managing KPIs in AIDA

With AIDA you can fine-tune KPIs prediction and retrain the prediction model after changes in the configuration. You can also access the KPIs Data Analysis page to analyze the KPIs trend over time.

About this task

From AIDA left-hand sidebar, select KPIs to open the KPIs page.

This page shows you the KPIs managed by AIDA.

From this page, depending on your permissions, you can run the following tasks:
  • Retrain the Machine Learning prediction model for one or more KPIs after changes in the configuration (for example, after adding a special day definition or changing a prediction tuning parameter).
    Note: Every 24 hours AIDA runs an automatic retrain of all KPIs. If you want the configuration changes to take immediate effect, you can request a retrain at any time.
  • Tune KPIs prediction parameters.
  • Access the KPIs Data Analysis page where you can:
    • Obtain an interval estimation of the KPIs trend.
    • Analyze the KPIs trend over time.
    • Identify and analyze anomalies in the KPIs trend.

The page header shows information about the latest planned retrain for all the KPIs (by default, automatic retrains run every 24 hours):

The search bar allows you to search for one ore more KPIs. The table displays the following information:
KPI Name
The name of the KPI.
Object Type
The type of the object measured by the KPI.
Object Name
The name of the object measured by the KPI.
Tag
A search tag for the KPI. Usually, the tag is used to identify the engine to which the KPI refers.
Anomaly %
The percentage of observed KPI data points that fall outside the expected range of values in the reference time interval:
  • < 6 : Low
  • 6-10: Medium
  • >10: High
Last Completed Retrain
Date and time of the last completed retraining process for the KPI.
Retrain Status
Status of the retraining process for the KPI. It can be: Completed, In Progress, Failed, Not enough data.

The Not enough data status means that there is not enough data available to generate a specific KPI prediction. Depending on the KPI, data needs to be collected from 1 to 7 days before the model can generate a prediction for the KPI.

Select one or more KPIs to run the following actions:
Retrain
To start the prediction model retraining process for the KPIs. KPIs Retrain cannot be scheduled by AIDA users. By default, the retraining process runs automatically on a daily basis (its start time is the start time of the Orchestrator container) , but you can start it upon needs (for example, after modifying a special day or a tuning parameter). This action is available to AIDA administrators only.
Tuning
To configure the KPI prediction parameters in the Tuning side panel. For example, you might run this action to increase or decrease the tolerance interval for anomaly detection. This action is available to AIDA administrators only. For details, see: Configuring AIDA for prediction.
Open
To open the KPI Data Analysis page where you can:
  • Obtain an interval estimation of the KPI trend.
  • Analyze the KPI trend over time.
  • Identify anomalies in the KPI trend.

For details, see: Analyzing KPIs data.

Notes:

KPIs cannot be delete.