Configuration

This section explains each of the properties in the configuration blade.

General

Property	Description
Collection	This associates the Agent to a specific Collection (default to that of the current Data Stream).

Refer to Collections and Stream Hosts to understand more about Collections.

Training

Property	Description
Learning Algorithm	The algorithm options used to train the model are: IID Change Point SSA Change Point IID Spike SSA Spike
Training File	The formatted file that contains data to train the learning algorithm.
Has Header	Tick to auto-populate the feature names from the first row of the training file, or the features are named Column1, Column2, etc.
Separator Character	The character options to separate data values in the training file are: Comma (,) Semicolon (;) Tab
Input Field	The training file attribute that provides training data to the algorithm.
Input Map	The column or attribute of the input payload that is examined by the algorithm. It must be of type Double.

Advanced Options

These options affect the training process of the Machine Learning algorithm.

Property	Description
Sensitivity	How sensitive the algorithm is to change between values. Accepts values between 0 and 100.
Change History Length	The size of the sliding window for computing the p-value (applies to Change Point Learning Algorithms only).
P-Value History Length	The size of the sliding window for computing the p-value (applies to Spike Learning Algorithms only).
Training Window Size	The number of points from the beginning of the sequence used for training (applies to SSA Learning Algorithms only).
Seasonality Window Size	An upper bound on the largest relevant seasonality in the input time series (applies to SSA Learning Algorithms only).
Spike Direction	Select whether positive, negative, or both positive and negative jumps in value will trigger an alert (applies to Spike Learning Algorithms only).

Model Options

Property	Description
Deterministic Seed	An optional positive whole number for deterministic results - i.e. the exact same output guaranteed for the exact same inputs across multiple runs.

Before configuring the AI & ML Agent, please ensure that its input endpoint is connected to a parent Agent which will be sending data to it.

Endpoints

Name	Description
Input	This endpoint is used to receive data from the parent Agent.
Output	Events received from the parent Agent are made available to this endpoint, with the results of the Machine Learning Algorithm appended to the original event: Alert, Score, PValue, and for some models, Martingale.