We and our partners use cookies to give you the best online experience, including to personalize content, advertising, and web analytics. You can reject cookies by changing your browser settings. To learn more about the cookies we use see our Cookie Policy.

Accept & Continue
AnyLogic is holding an in-depth offline training course in Lisbon, 25-27 of March. Learn from the best!
AnyLogic Training Course in Lisbon
Learn from the best!
AnyLogic

Enhancing the Predictive Capabilities of Simulation with H2O.ai Automatic Machine Learning

AnyLogic is proud to partner with H2O.ai, the leading automatic machine learning platform. Now, you can use the distinct predictive capabilities of simulation and machine learning together.

To get started with incorporating machine learning (ML) models trained with H2O.ai Driverless AI, we have two documented proof-of-concepts models for download. Embedding ML models into simulations enables new areas of application: data scientists can test solutions in a risk-free space and simulation modelers can access more data-driven inputs. Read on, learn more, and try it for yourself!

Machine learning (ML) solutions are increasingly becoming mainstream in business. And as more of them are deployed and integrated with existing systems, both the data science and simulation communities using them can benefit from each other’s distinct capabilities.

For simulation modelers:

  • The work in building simulation models has always involved improving the accuracy of the virtual environment as it relates to the real system. Without direct access to the causal rules that govern a real system, it is necessary to approximate the outcome of different scenarios with all types of probabilistic and statistical models; ML models are the latest advancements in the evolution of these data driven components. Simulation models should make use of these newer breed of predictive models, as more of them become available as part of data-oriented investments in business.
  • In replicating the rules of a real system, rules and behaviors that are a direct result of a system’s embedded AI solutions should also be incorporated in the simulation. The most natural way of achieving this is to directly embed the AI solutions into the simulation.

For data scientists and AI experts:

  • The objective of adding AI components to a system is to improve overall system performance, not just of the specific components being substituted by AI. It is a reasonable expectation that deploying a well-trained AI solution will have a significant improvement on the overall performance of the target system. However, any perturbation in a system has the potential to shift any bottlenecks or cause other ripple effects on the system. Testing a trained model on its own does not verify that the performance of the modified system (as a whole) is sufficiently improved. Simulation models can be used as a virtual, risk-free environment to test the implications of incorporating AI into existing systems.
H2O.ai platform

H2O.ai Driverless AI Scoring Pipeline

H2O Driverless AI offers [ML] model deployment, management and monitoring capabilities for IT and DevOps teams. In machine learning, pipelines are the automation of sequential steps in a workflow. These steps may include data preparation, [ML] model training, validation, packaging, and deployment as well as monitoring. A scoring pipeline is usually a part of the deployment routine where trained models are used to make predictions on new data.


AnyLogic H2O.ai platform workflow

For embedding trained ML models into AnyLogic, Driverless AI provides a low-latency standalone Model Object which can be scored in real time: the Optimized (MOJO) Scoring Pipeline. This trained ML model is downloadable as a standalone file that can be incorporated into AnyLogic models. It can then be used like a function that returns the desired prediction based on the inputs that are dynamically passed from the simulation model during runtime.


Webinar: Combining Simulation and Machine Learning

H2O Driverless AI automates time-consuming ML tasks so that data scientists can work faster and more efficiently. Automated tasks include model validation, model tuning, model selection, and feature engineering.

This webinar shows how the different predictive abilities of simulation and machine learning combine to advance decision support in business and public enterprise. Arash Mahdavi, AnyLogic AI Program Lead, is joined by Data Scientist Niki Athanasiadou and Senior Solution Architect Heman Kapadia from H2O.ai. With an example model, they demonstrate how to improve its predictive capability by embedding the H2O Driverless AI MOJO pipeline.

Example models


Here are two AnyLogic example models that were refactored for the H2O.ai platform. They are already set up and ready to go, just download them from the links below. Comprehensive documentation is also provided in their companion README.md file.

  • 01

    Hospital Capacity Planning

    Hospital Capacity Planning model

    This example model demonstrates a use-case for capacity planning and management in hospitals that deal with unprecedented surges in patients (e.g., COVID-19). In this model, a hospital is simulated where patients occupy resources (hospital beds) for a certain time duration; the specific amount of time spent is predicted by an embedded Driverless AI MOJO pipeline based on each incoming patient’s attributes and preexisting conditions. The arrival rate of patients and the resource capacity (total beds) can be dynamically changed via the provided controls. In this way, various scenarios can be tested to check if the set capacity can cope with the arrival rate.

    Download

  • 02

    Product Delivery

    Product Delivery model

    This example model demonstrates a supply chain that operates based on fluctuating local and global demand variables. There are fifteen distributors that order products every 1 to 2 days. The amount ordered is based on both seasonality and whether there is an active pandemic. Three manufacturing facilities fulfil orders by dispatching either from inventory or by waiting for production.

    The model uses two embedded Driverless AI MOJOs. The first predicts the current temperature of each distribution center based on its location and current date. The second predicts the demand on each distributor based on location, date, predicted temperature (from the first embedded Driverless AI model), and pandemic presence. The predictive capabilities of the two embedded ML models allow us to simulate and dynamically monitor the overall performance of the supply chain with and without the presence of a pandemic.

    Download

Simulation for training and testing AI – Email Pack

AnyLogic simulation is the training and testing platform for AI in business. With AnyLogic general-purpose simulation, you can construct detailed and robust virtual environments for training and testing your AI models. The unique multi-method simulation capabilities provide a comprehensive tool for use in machine learning. Established in use at leading companies across industries, this fully cloud enabled platform with open API is enhancing and accelerating AI development today. Find out more about this powerful machine learning tool in our AI email pack and white paper!

AI pack and white paper