Splunk® User Behavior Analytics

Administer Splunk User Behavior Analytics

Acrobat logo Download manual as PDF


This documentation does not apply to the most recent version of Splunk® User Behavior Analytics. For documentation on the most recent version, go to the latest release.
Acrobat logo Download topic as PDF

Customize anomaly scoring rules

You can customize the scoring rules for anomalies in Splunk UBA.

Anomaly scores are derived from a variety of factors and conditions defined in a rule. Customizing the scoring rules means you can provide a level of consistency in how anomalies in your environment are scored. For example, the Excessive Data Transmission anomaly is triggered based on the following rules:

  • Analysis of Baseline
  • Analysis of Peer Group
  • Analysis of Transfer

As an example, you can change the scoring rules for this anomaly by elevating the threshold for the Analysis of Peer Group rule and generating a higher anomaly score only when the threshold for this rule is exceeded. This particular rule is also used to score other anomalies, like Excessive Box Downloads. By making the same changes to this rule for all anomalies, you can create consistency in your environment in terms of which factors are emphasized to generate the anomalies and threats you want to see.

Viewing the scoring rules also provides insight into the constitution of each anomaly, and help you understand how to modify the scoring rules based on your needs.

You can also change the scoring ranges for the anomalies themselves, to emphasize certain types of anomalies over others. For example, you may change all anomalies related to data exfiltration to have a higher maximum score than any other anomalies in your system to ensure that threats generated by data exfiltration anomalies will also have higher scores and be prioritized accordingly.

A summary of each anomaly's scoring rules can be viewed in the Score Details pane on the Anomaly Details page. See Review the anomaly details.

View and understand anomaly scoring rules

To view the scoring rules associated with each anomaly, perform the following tasks:

  1. In Splunk UBA, select Anomalies from the home page, or select Explore > Anomalies from the menu bar.
  2. Click the anomaly rules (the gear, or anomaly rules icon) icon.
  3. Click Anomaly Scoring Rules.
  4. Select an anomaly from the list of anomalies to view the scoring rules.

Click on each anomaly name to view the scoring rules for that anomaly. The example below shows the Suspicious Network Exploration anomaly:

This screen image shows the scoring rules for the Suspicious Network Exploration anomaly. The main items in the screenshot are described in the following text.

View the scoring rules for this anomaly in the Anomaly Scoring Rules section. Scores for the Suspicious Network Exploration anomaly are based on the following rules:

  • Event Ids Analysis
  • Expansion Analysis
  • User Account and Device Features Analysis

Each rule listed also contains a brief description about the rule. The "No" in the Customized column means that the scoring rule has not been edited and is the same as originally shipped with Splunk UBA.

View the base scores for the anomaly in the Anomaly Base Score section. Every anomaly consists of a base, minimum, and maximum score value. These are described in the following table.

Score Type Description
Base Score Every anomaly has a base score, which is derived from analyzing the events in your system to find anomalous indicators. Scoring rules add additional context and related metadata to the base score.


In general, anomalies like Blacklisted Application that do not have a lot of rules and additional metadata associated with it tend to have higher base scores, while anomalies like Suspicious Network Connection that include multiple rules tend so start with lower base scores. It is possible for some anomalies to have a base score that is lower than the minimum score.
You can have different base scores for the same type of anomaly. For example, two separate Suspicious Network Connection anomalies can have different base scores due to factors such as the number of connections, deviations from the baseline, and peer groups.

Min Score Lowest score that this anomaly can have.
Max Score Highest score that this anomaly can have.

The final anomaly score is a value between the minimum score and the maximum score.

Customize the anomaly scoring rules

To customize the scoring rules for an anomaly, perform the following tasks:

  1. In Splunk UBA, select Anomalies from the home page, or select Explore > Anomalies from the menu bar.
  2. Click the anomaly rules (the gear, or anomaly rules icon) icon.
  3. Click Anomaly Scoring Rules.
  4. Select an anomaly from the list of anomalies to view the scoring rules.
  5. Hover over the rule name on the Anomaly Scoring Rules page and click on the edit (the pencil, or edit icon) icon.

The example below shows the details for the Event Ids Analysis scoring rule.

This screen image shows the Edit Anomaly Scoring Rule window. The items in the window are described in the text immediately following this screen image.

The items on this page are described in the following table:

Column Description
Factor Name and brief description of the factor contributing to the rule.
Conditions The thresholds that determine when the score for the anomaly is changed. This field contains a variety of formats, depending on the nature of the condition. In this example, we can make the following changes:
  • Change the threshold of the Presence of Abnormal Event Id Frequency factor by selecting = as the operator and entering 1.
  • Change the threshold of the Many Changes in Event Ids Frequency factor by selecting > as the operator and entering 5.
Score Delta The amount to increase or decrease the score if the specified condition is met.
  • In the Presence of Abnormal Event Id Frequency factor, leave the score delta as 1.
  • In the Many Changes in Event Ids Frequency factor, change the score delta to 2.

In cases with multiple scores, there is an implied AND condition. In this example, if both Presence of Abnormal Event Id Frequency and Many Changes in Event Ids Frequency are triggered, the anomaly score is incremented by 3. You can disable the Presence of Abnormal Event Id Frequency by deselecting the checkbox in the Enabled column, so that when the Many Changes in Event Ids Frequency threshold is crossed, the anomaly score is incremented by 2.

Enabled Make sure the checkbox is selected to have this rule condition be active. Deselect it to make the condition inactive.

The image shows what the rule looks like after the changes.

This screen image shows the Edit Anomaly Scoring Rule window. The items in the window are described in the text immediately above this screen image.

Customize the anomaly base, minimum, and maximum scores

To customize scoring for the anomaly, perform the following procedure:

  1. In Splunk UBA, select Anomalies from the home page, or select Explore > Anomalies from the menu bar.
  2. Click the anomaly rules (the gear, or anomaly rules icon) icon.
  3. Click Anomaly Scoring Rules.
  4. Select an anomaly from the list of anomalies to view the scoring rules.
  5. Hover over the row with the anomaly scores and click the edit (the pencil, or edit icon) icon.
  6. In the Edit Anomaly Base Scores window, change the anomaly scores as desired.
  7. Click OK.

Recomputing anomaly scoring rules

After you finish editing the anomaly scoring rules, click Recompute Anomaly Scores on the Anomaly Rules page to have your changes take effect. This can take up to several hours, depending on the number of anomalies in your system. Splunk UBA maintains an internal versioning system for all anomaly rules that determines which version of a rule to use when recomputing anomaly scores after edits are made, or Splunk UBA is upgraded.

The internal versioning system contains a major and minor version number.

  • The major version number is updated as needed whenever Splunk UBA is upgraded
  • The minor version number is updated whenever a scoring rule is edited.

All anomalies generated and added to the system will use the scoring rules of the latest major and minor version. For example, when a Suspicious Network Exploration anomaly is generated, it will be scored based on the latest major and minor version of the scoring rule for this anomaly.

The following table summarizes what happens to the existing anomaly scores in your system when you make changes to a scoring rule, or upgrade Splunk UBA.

Event Impact on anomaly scores
You edit an anomaly scoring rule. The minor version number for that anomaly is updated. All past anomalies with the same major version and lower minor version number are recomputed, except for anomalies that were scored by an anomaly action rule which is still an active rule. If there are existing anomalies of this type that use an older scoring model (earlier than Splunk UBA 4.2.0), those anomalies are not recomputed. In addition, any new anomalies generated based on the old scoring model are also not re-scored.
You upgrade Splunk UBA. If a scoring rule for a particular anomaly is updated as part of a Splunk UBA release, then the major version number for that anomaly is updated. All customizations for previous versions of the same anomaly are lost.

Reverting anomaly score rules

You can revert an anomaly scoring rule back to the Splunk UBA default values. If you choose to do this, all of your customizations will be lost.

To revert an anomaly scoring rule back to the Splunk UBA default:

  1. From the anomaly scoring rules page, hover over the scoring rule you want to revert and click on the revert (the revert icon) icon.
  2. Click OK to confirm that you want to revert all customizations made to this rule.
Last modified on 12 December, 2022
PREVIOUS
Take action on anomalies with anomaly action rules
  NEXT
Mask personally-identifiable information in Splunk UBA

This documentation applies to the following versions of Splunk® User Behavior Analytics: 5.0.0, 5.0.1, 5.0.2, 5.0.3, 5.0.4, 5.0.4.1, 5.0.5, 5.0.5.1, 5.1.0, 5.1.0.1


Was this documentation topic helpful?


You must be logged into splunk.com in order to post comments. Log in now.

Please try to keep this discussion focused on the content covered in this documentation topic. If you have a more general question about Splunk functionality or are experiencing a difficulty with Splunk, consider posting a question to Splunkbase Answers.

0 out of 1000 Characters