Initial Setup

Our AI Anomaly Detection System is fully customizable, and gives your plenty of features in term of rule configuration and AI Model builds, it comes with a MLFlow system out of the box, which gives you the ability to build your AI model based on the rule detection as well as the data ingested in the system.

Step 1

Create your organization:

Once sign in, the first step is the organization setup, you can click on setup organization to be redirected to the configuration form. two information are needed organization name and an external Id.

Submitting the requests will generate an API Key which can be used while calling the verification API, it should be passed as header using X-API-Key. Please make sure that this information is not shared since it may impact your billing.

Step 2

Setup Your organization Info

Follow this simple form to set to initial information needed, once submitted an API-KEY will be generated and can be used to reach our backend system

Step 3

Schema Request Configuration:

You have two options, either upload a sample json request, or copy and paste a sample you already have.

The request Key play the role of the operation key, e.g: Account Number, Customer Identifier…. Then You need to specify the PATH of that key in the request, currently we support a simplified form such us /store/book

Since most of the anomaly operation checks are made against a period of time, you can specify that the date are included in the request and the system will rely on it instead of the generated value

Step 4

Rule Anomaly Configuration

Once Your rule has been configured with the necessary options, criteria, aggregates and the actions to fire. you can make view your configuration in Settings->Schemas.

Once everything is set you’ll be redirected to the configured requests screen, at this level you can activate the AI dynamic build.

Step 5

AI Activation

By enabling AI, the model can be built either from the Machine Learning section or dynamically by the system.

The scoring system will start returning the prediction as well as the score, please note that the system needs a minimum level of data to have an accurate AI prediction model, so please make sure to not take the scoring into consideration unless the request model has been built with a good amount of data and that the accuracy is at least 80%.

Step 6

AI Attributes

AI Anomaly Detection relies on the configured and selected attributes. So to be able to benefit from the machine learning detection system, you need to set the attributes as well as the type of each one, this will be the base of the model build. The system will make the necessary transformation based on the features types before building the model.