Introduction

Learning-Based First Response uses machine learning to predict alert suppression and snooze actions based on historical alert behavior. The model learns recurring alert patterns and recommends automated first-response actions before incident creation or escalation occurs.

Access the First Response ML Policy

1. Navigate to Setup > Account > Alert Policies.
2. The Command Center AIOps Overview page is displayed.
3. In the Policy Types section, click First Response.
4. On the Alert Policies page, click Machine Learning.

The Machine Learning - Alert Policies page opens with the First Response tab selected.

Understand the First Response Training Dataset

The training dataset contains Input and Output columns.

Input Columns

Input columns capture the alert attributes that the model uses to detect repeatable patterns.

Examples include:

• metric
• resource.generalInfo.resourceType
• currentState
• component

Sample values are shown below.

Output Columns

The model predicts the first-response actions to apply.

Examples include:

• suppressed
• snoozeDuration

Sample values are shown below.

The model uses these mappings to determine whether alerts should be automatically suppressed or snoozed and for how long.

Add Training Records

1. Click Add.
2. Enter values for the Input fields.
3. Enter the corresponding suppression and snooze values.
4. Save the record.

Search Training Data

Use the Search icon to locate records within the dataset.

The search returns matching rows and displays the number of results found.

Filter Data Within a Column

Each Input and Output column supports value-level filtering.

To filter:

1. Click the filter icon beside a column name.
2. Search for available values.
3. Select one or more values.
4. The table displays only matching records.

Apply Advanced Filters

Use the Filter option to filter records across multiple columns simultaneously.

Examples include the following fields.

Input Filters

• metric
• resource.generalInfo.resourceType
• currentState
• component

Output Filters

• suppressed
• snoozeDuration

Configure Input and Output Columns

Click the Settings icon to define the fields used during training.

You can:

• Add new Input columns.
• Enable or disable Input columns.
• Configure Output columns.
• Exclude unused attributes from model training.

A maximum of 15 Input columns can be configured.

Seasonal Patterns

The First Response policy displays a Seasonal Patterns (SP) indicator.

This indicator shows the status of seasonal pattern detection used by the model.

For example:

Seasonal Patterns: AWAITING DATA

The status indicates that additional data is required before seasonal patterns can be identified.

Import and Export Training Data

From the More Actions menu (…), select:

• Import CSV Training File
• Export Table as CSV

Train the Model

1. Review the training records.
2. Click Save and Train.

The platform trains the First Response model and updates the Last trained time displayed on the page.