search

IBM / ibm-maximo-contextual-anomaly-detection-models

this is for setting up tutorial on contextual anomaly detection work. The tutorial cover a sample illustration of the API.
Apache License 2.0
1 stars 0 forks source link

readme

Contextual Anomaly Detection Accelerator Guide

Overview

The contextual anomaly model is designed to handle multivariate data as input and predicts both a lower and upper value that encompasses the target variable/signal under normal conditions, based on a user-specified probability or confidence level. The higher the chosen probability, the less likely it is to observe the monitored signal outside of the predicted interval under in-distribution conditions. Consequently, any measured target variable falling outside the predicted interval can be considered an anomaly, with the probability of false alarm equal to the probability specified for the prediction interval.

Two typical examples:

API Documentation

API Documentation is hosted on Developer Hub

Onboarding Guide

This repository includes a comprehensive guide for building and deploying a conformal prediction based model using Contextual Anomaly Recipe (available as API endpoint) for Maximo Monitor Application. It provides all the necessary resources, including

  1. Sample data for Wind Turbine Asset
  2. AI Cookbook to facilitate model training using API Service,
  3. Creating an IoT device on Monitor,
  4. Deploying and configuring the KPI endpoints to the created device,
  5. Conducting inference with real-time data streaming into Monitor,
  6. Visualizing the results through an intuitive dashboard

Navigate to the onboarding folder, where you'll discover all the resources essential for utilizing our contextual anomaly detection accelerator. To streamline and automate the entire procedure, we have provided a detailed outline of the required steps.

Completing the walkthrough should take around 15 minutes, taking into account the time needed for training, deployment, and inference on a dataset of size 10,000 during the demonstration.

Security

See CONTRIBUTING for more information.

License

This collection of AI cookbooks is licensed under Apache. See the LICENSE file.

Questions

Please raise an issue on this repository.