TMIP-EMAT documentation

Warning

The views expressed in this documentation do not necessarily represent the opinions of FHWA, and do not constitute an endorsement, recommendation, or specification by FHWA.

TMIP-EMAT is a methodological approach to exploratory modeling and analysis. It provides a window to rigorous analytical methods for handling uncertainty and making well informed decisions using travel forecasting models of all types. It is designed to integrate with an existing transportation model or tool to perform exploratory analysis of a range of possible scenarios.

TMIP-EMAT provides the following features to enhance the functionality of the underlying core model:

• A structure to formalize and distill an exploratory scope, in a manner suitable for translating the abstraction of the “XLRM” robust decision making framework into a concrete, application-specific form,
• A systematic process for designing experiments to be evaluated using the core model, and support for running those experiments in an automated fashion.
• A database structure to organize and store the results from a large number of experiments.
• A facility to automatically create a metamodel from experimental results, which uses machine learning techniques to rapidly approximate new outputs of the core model for without actually running it for every relevant combination of inputs.
• A suite a analytical and visualization tools to explore the relationships between modeled inputs and outputs, and develop robust policy strategies that might be effective across a range of possible future scenarios.

The methodologies demonstrated in this toolset build on and extend the capabilities of the EMA Workbench, version 2. A brief discussion of the differences between these projects is available.

For an introduction to the concepts and tools contained in TMIP-EMAT, you can view our TMIP Webinars and Screencasts:

• Introducing the Exploratory Modeling and Analysis Tool
• Using TMIP-EMAT for Exploratory Analysis
• TMIP-EMAT: Setting up a Model
• TMIP-EMAT: Running a Model
• TMIP-EMAT: Scatter Plot Matrixes and Feature Scoring
• How to Get Started with TMIP-EMAT
• Using TMIP-EMAT for Decision Making under Deep Uncertainty
• Connecting TMIP-EMAT and VisionEval for Exploratory Analysis

In addition to this documentation, there are several helpful example repositories hosted on GitHub:

• TMIP-EMAT Source The main repository for TMIP-EMAT source code that demonstrates various aspects of the exploratory modeling methodological approach.
• Using TMIP-EMAT with VisionEval This repository demonstrates the use of TMIP-EMAT with the VisionEval’s Regional Strategic Planning Model (RSPM).
• Using TMIP-EMAT with a Bespoke External Model This repository demonstrates connecting TMIP-EMAT with an arbitrary custom model that is controlled by configuration files and invoked on the command line. A similar approach can be used to connect TMIP-EMAT to most commercial and proprietary travel demand models.

TMIP-EMAT is provided under a BSD 3-Clause License license. It also includes similarly licensed code from the EMA Workbench.

Contents:

• Introduction to EMAT
  • The Core Model
  • End User Requirements
  • Model Inputs, Outputs, and Configurations
    • Inputs
    • Outputs
    • Configurations
• Installation and Configuration
  • Installing using Anaconda
    • Quick Start
    • Installing Python
    • Managing Environments
    • Running Jupyter
    • Troubleshooting
  • Installing from Source
    • Using Github to Fork TMIP-EMAT
• Exploratory Scoping
  • Scope
    • Read / Write
    • Feature Access
    • Utilities
  • Parameters
    • Float-Valued Parameters
    • Integer Parameters
    • Boolean Parameters
    • Categorical Parameters
  • Performance Measures
  • Boxes
    • Box
    • Boxes
    • ChainedBox
  • Scope File
    • Scope Name and Description
    • Inputs
    • Outputs
• Experimental Designs
  • Univariate Sensitivity Tests
  • Efficient Designs
  • Reference Designs
  • Experimental Design API
• Core Models
  • Basic EMAT Model API
    • Abstract Methods
    • Data Management
    • Model Execution
    • Meta-Model Construction
  • Python-Based Models
  • Excel-Based Models
  • Files-Based Model
    • Writing and Using a Bespoke Model Interface
    • TableParser Example
    • MappingParser Example
    • Files-Based Model API
  • GBNRTC Model
    • Implementation-Specific Methods
• Meta Models
  • Meta-Model Creation
    • Design Sizing
    • Partial Metamodels
  • Meta-Model API
  • Meta-Model Regression
• Databases
  • Basic EMAT Database API
    • Scopes
    • Boxes
    • Experiments
    • Meta-Models
  • SQLite Database
    • Scopes
    • Boxes
    • Experiments
    • Meta-Models
  • Database Walkthrough
    • Single-Design Datasets
    • Multiple-Design Datasets
    • Re-running Experiments
• Analysis
  • Visualization
    • Scatter Plot Matrix
    • Identifying Model Problems
    • Interactive Visualizer
  • Scenario Discovery
    • Feature Scoring
    • PRIM
    • CART
  • Directed Search
    • Policy Contrast
    • Optimization Tools
    • Mechanics of Using Optimization
• Examples
  • Road Test
    • Defining the Exploratory Scope
    • Using a Database
    • Experimental Design
    • Core Model in Python
    • Feature Scoring
    • Visualization
    • Scenario Discovery
    • Robust Search
    • Creating a Meta-Model
  • GBNRTC Example Model
• Glossary
• Methodology Index
• References

Indices and tables

• Index
• Module Index
• Search Page