Using offloaded analytics data to train AI models
How to use API Connect analytics data to train AI models.
API Connect analytics holds full information on all of your API events. This information can prove invaluable in training AI models by using machine learning. Possible use cases include training data for automated tests or for security anomaly detection.
IBM Cloud Pak® for Data (CP4D) includes the means to train AI models with data from various data sources. It is not possible for CP4D to pull data directly from API Connect's internal analytics database because the database is secured with no external access. However, you can offload your analytics data to an external database, and then configure CP4D to load data from that external database. You can use any external database that both API Connect analytics offload and CP4D support. The simplest option might be an external ELK stack (Elasticsearch Logstash Kibana). A list of supported data sources for CP4D can be found here: https://www.ibm.com/docs/en/cloud-paks/cp-data/latest?topic=data-supported-sources.
Documentation on how to configure Cloud Pak for Data to access Elasticsearch can be found here: https://www.ibm.com/docs/en/cloud-paks/cp-data/latest?topic=catalogs-elasticsearch-connection.
When your analytics data is offloaded to an external database, you can use it to train your AI model. Here is an example of how to create and train an AI model in CP4D using IBM Watson® Machine Learning: https://www.ibm.com/docs/en/cloud-paks/cp-data/latest?topic=deploying-managing-ai-assets.