BookmarkSubscribeRSS Feed
☑ This topic is solved. Need further help from the community? Please sign in and ask a new question.
agesser
Obsidian | Level 7

HI all,

There is option to add Priority Score column to the Alert's grid.

How we can know what is the calculation behind these scores? Can we have any documentation to understand how the calculation is working?

 

Thanks in advance.

Alex.

1 ACCEPTED SOLUTION

Accepted Solutions
FMinago
Fluorite | Level 6

Hi

Please refer to the chapter Alert Scoring in SAS AML 8.3 documentation . 

 

Additionaly you can explore the code in Alert encirchment (fcf_aml_alert_enrich.sas) and Alert scoring (fcf_aml_alert_scoring.sas) JES jobs

https://documentation.sas.com/doc/en/compcdc/8.3/amlag/ch-alert-scoring.htm

The following steps highlight the overall scoring process for the out-of-the-box model:

  1. Both historical enriched data and active enriched data are appended together.
  2. All missing values are imputed using the mean of the observed values for each variable.
  3. All variables are standardized by subtracting a location measure and dividing by a scale measure.
  4. An autoencoder is used to compress the high dimensional feature space to a 2D space.
  5. Clusters corresponding to the productive dispositioned alerts are identified in the 2D space.
  6. Scores are assigned by means of the Distance to the nearest productive alert cluster.
  7. A reason code is generated to explain the score.

View solution in original post

3 REPLIES 3
SASKiwi
PROC Star

Documentation for AML is locked down so it can only be accessed by AML customers. Asking this in a Tech Support track is probably a better option.

FMinago
Fluorite | Level 6

Hi

Please refer to the chapter Alert Scoring in SAS AML 8.3 documentation . 

 

Additionaly you can explore the code in Alert encirchment (fcf_aml_alert_enrich.sas) and Alert scoring (fcf_aml_alert_scoring.sas) JES jobs

https://documentation.sas.com/doc/en/compcdc/8.3/amlag/ch-alert-scoring.htm

The following steps highlight the overall scoring process for the out-of-the-box model:

  1. Both historical enriched data and active enriched data are appended together.
  2. All missing values are imputed using the mean of the observed values for each variable.
  3. All variables are standardized by subtracting a location measure and dividing by a scale measure.
  4. An autoencoder is used to compress the high dimensional feature space to a 2D space.
  5. Clusters corresponding to the productive dispositioned alerts are identified in the 2D space.
  6. Scores are assigned by means of the Distance to the nearest productive alert cluster.
  7. A reason code is generated to explain the score.
agesser
Obsidian | Level 7

Thanks @FMinago