Kolmogorov.ai Predicate
Updated February 10, 2025
Kolmogorov.ai Predicate — an application for performing monitoring or quantitative validation, which enables the organization of a monitoring project based on a vast library of metrics and allows the creation of a validation report from multiple projects.
Configure monitoring projects from statistical and business metrics. The Predicate Metrics Catalog contains a wide array of built-in metrics, as well as allows for the addition of custom calculation algorithms for indicators.
Predicate provides the following functionality:
- A catalog of validation and monitoring projects.
- A catalog of metrics with Plotly visualizations, addition of custom Python metrics.
- A pre-configured package of standard model quality assessment metrics (including data / concept drift), as well as metrics specific to subject areas like credit risk, operational risk, and customer analytics.
- Scheduled project execution according to a set regimen.
- A catalog of template projects for standard monitoring and validation.
- A GUI for constructing validation reports.
- The ability to add custom preprocessing code to the execution of monitoring/validation projects (for example, ad-hoc model application, factor calculation, etc.).
1. Prerequisites
Components that are not included in the default delivery of the Predicate application but are required for its operation:
- A
Kubernetes
-compatible cluster with the ability to create a separate namespace. In this namespace — the ability to create a service account with CRUD rights for all basic types of objects. Minimum resource quota —8CPU, 16GB RAM, 100GB disk space
. Additionally, the following components are required in the cluster:NFS StorageClass
— RWM storage class in Kubernetes that uses an NFS server.Ingress Controller
— a controller that provides access to Kubernetes services from outside.
Keycloak
— a service for authorization and authentication using Keycloak.S3 storage
(Minio or Yandex Object Storage) — storage where temporary files and metric results are saved. In this storage, there should be the ability to create a separate bucket. In this bucket — a service account with the role ofstorage.editor
orstorage.admin
(GET/UPDATE/DELETE objects).#### 2. Installation and Configuration of Components
Installation and configuration of components and cluster
2.Instruction for checking Predicate’s operability.
3.- Rapid generation of modeling results reports for communication with key project participants and stakeholders.
- Automation of quantitative validation tasks.
- Operational dashboards for scheduled monitoring of models and solutions in an industrial format.
- Standardization and templating of monitoring regulations for diverse scenarios.
- Setting up notifications and triggers for automatic calibration and retraining based on a flexible traffic light system.
Helm chart | Version | Pull-command | Documentation |
---|---|---|---|
datasapience/kolmogorovai-predicate/charts/predicate | 2.1.2 | Open |
Docker image | Version | Pull-command |
---|---|---|
datasapience/kolmogorovai-predicate/predicate1739201023865532027990309722966592018714261469144 | 2.1.2 | |
datasapience/kolmogorovai-predicate/predicate-ui1739201023865532027990309722966592018714261469144 | 2.1.2 | |
datasapience/kolmogorovai-predicate/bitnami-rabbitmq1739201023865532027990309722966592018714261469144 | 3.13 | |
datasapience/kolmogorovai-predicate/bitnami-postgresql1739201023865532027990309722966592018714261469144 | 14 |
By using this product you agree to the Yandex Cloud Marketplace Terms of Service