Machine Learning Overview

Machine Learning Overview

Rayven’s combined AI + IoT platform does everything within a single system. Machine learning enables the real-time optimization of workflows and solutions created elsewhere in the system.

The machine learning workbench utilizes the same drag-and-drop logic as the workflow builder to ensure users of any ability can create and add sophisticated predictive analytics to their real-time monitoring solutions.

Rayven provides 3 ways of utilizing the machine learning workbench: 

1. Use one of the Rayven ready-to-go data models.
2. Create your own data model using the Rayven platform. 
3. Upload your own Python-based data model.
   

A Fully-integrated AI + IoT solution

Rayven's machine learning tool kit integrates fully with the rest of the real-time IoT platform. This connectivity ensures no data gaps, less overall complexity, and better results.

Adding Machine Learning to your solution

The AI Dynamix predictive analytics modeler and engine uses a simple drag-and-drop interface for data, actions, and visualizations.

Using the same technology as the Workflow Builder, you can integrate and collect data from any system when creating your real-time IoT monitoring and management solution.

Implementing Machine Learning is an easy 6-step process:

1. First, prepare your data. Select training data from your workflow and apply ready-to-go filters and normalizers from the built-in Machine Learning Workbench options. 
2. Next, build or choose your algorithm. Create an algorithm or import a Python algorithm off-the-shelf for your use case. Rayven has ready-to-go algorithms for a wide variety of uses, including forecasting, anomaly detection, predictive maintenance, and more.
3. Train your algorithm by applying it to prepared offline data. Leave it to create a model that can answer the questions you pose to it.
4. Once you have trained your chosen model, drop it into the Workflow Builder using a Machine Learning node to add predictive analytics to your existing solution. 
5. Over time, you (and your algorithm) will learn from missed breakdowns and evolve the model's sophistication. Through AI Dynamix, you can easily make adjustments and improvements.
6. Compare and contrast algorithms by running multiple models simultaneously and testing which works best for you. The platform can run concurrent models for different purposes, too.