Performance Metrics & Data Analytics

Services Provided: Artificial Intelligence, Travel Demand Modeling

Insight is part of a team that is pioneering an initiative with the Florida Department of Transportation’s Forecasting & Trends Office (FTO) to integrate AI into travel demand modeling. The project is part of FDOT’s broader efforts to integrate innovative technologies into planning workflows to enhance model reliability, efficiency, and reproducibility. Specifically, Insight is supporting the development of an AI-driven run routine designed to automate and enhance the calibration and validation of travel demand models, beginning with the Florida Hurricane Evacuation Model, also known as TIME. 

The primary objective of this task is to increase the accuracy and consistency of model outputs while minimizing manual intervention and the potential for human error. The project aims to deliver a deployable routine that incorporates artificial intelligence (AI) methodologies into the model calibration and validation process. Insight’s high-level contributions include AI application development, model calibration and validation automation, and system integration, ensuring compatibility with all district and statewide planning models used by FDOT. 

Insight is currently developing guidance for implementing AI solutions and identifying the value of open-source versus proprietary AI options. Subsequent efforts will focus on designing an AI-driven auto-calibration framework for trip distribution in the TIME model, incorporating optimization algorithms that adapt to shifts in behavior and infrastructure conditions. Further work involves drafting a comprehensive user manual, developing training materials, and leading webinars to facilitate statewide adoption. Insight also contributed to broader support activities such as data gathering, trend evaluation, meeting facilitation, and the development of materials for stakeholder communication.  

The work will culminate in the delivery of a complete AI-enhanced model run routine and associated documentation by the end of 2026.

Services Provided: Survey Expansion, Travel Market Analysis, Incremental STOPS Model Development   

Insight has served as a subconsultant to the ETC Institute for numerous Systemwide Onboard Transit Surveys across the country since 2022. Insight has supported on-board survey collection in Denver, Columbus, Hampton Roads, Broward County, Minneapolis-St. Paul, Houston, Los Angeles, and Inglewood. The overarching goal of these projects is to provide high-quality ridership data that would directly inform regional travel patterns and enhance the regional travel demand modeling and forecasting capabilities. 

Insight’s role has been to provide methodological support for survey implementation and analytical enhancements to surveys, ensuring that the survey data is robust. Insight identifies the essential variables needed from the transit surveys for the regional transit ridership forecasting models using the Federal Transit Administration’s (FTA) Simplified Trips on Project Software (STOPS). For each survey, Insight efforts are focused on the reconciliation of the primary expansion with the APC counts and performing a secondary expansion of the survey (ETC performs the primary expansion). The secondary expansion enhances the value of the dataset by incorporating access/egress modes, transfer activity, and rider demographics. Insight’s work ensures the final survey-based dataset meets the region’s modeling and planning needs. Insight also incorporates the survey into the existing STOPS model and provides high-level recommendations on additional updates to the STOPS model. 

Services Provided: Traffic Forecast Data Analysis, Statistical Analysis

Insight is leading an effort for the Metropolitan Council to evaluate the accuracy and robustness of long-range traffic forecasts. This project builds on decades of traffic forecasting research and marks the third regional forecast accuracy assessment conducted in the Twin Cities, and the first to incorporate the methodological advancements from the National Cooperative Highway Research Program (NCHRP) Report 934.

The objective of this work is to assess the reliability of past traffic forecasts, improve modeling procedures, and develop tools to better communicate forecast uncertainty to planners and decision-makers. The analysis will enhance both the Met Council’s and Minnesota Department of Transportation’s (MnDOT) understanding of how forecasted and actual traffic volumes have aligned over time, providing a critical foundation for future infrastructure planning. The key activities are described below: 

• Compiling a robust database of past traffic forecasts, traffic counts, and project metadata. Insight created an innovative ChatGPT- and Python-powered tool to extract key forecast data from over 450 MnDOT traffic reports—many of them lengthy and scanned—into a structured, tabular format.  
• Applying multiple statistical techniques—such as percent difference from forecast (PDFF), quantile regression, and longitudinal comparisons—to quantify forecast accuracy. This analysis will be compared against prior regional assessments to measure improvement over time. 
• Providing recommendations to improve both forecasting procedures and how forecast uncertainty is communicated. Insight will leverage quantile regression and user-focused visualization methods to support decisions under uncertainty and will suggest refinements to the Met Council’s travel demand model based on systematic errors identified in the forecast database.

This study provides the Met Council with a highly defensible and transparent record of forecast accuracy, strengthens the credibility of travel demand models in future decision-making, and equips planners with tools to explain and manage forecast uncertainty in public and interagency settings. Insight’s leadership in this project showcases its unmatched expertise in travel demand model evaluation and its ability to merge technical rigor with actionable, policy-relevant recommendations. The methodologies and tools developed will be scalable for use by MPOs and DOTs nationwide.