Project Portfolio

In their short time as PhD students at the University of Colorado Boulder, Matt and Guillermo worked on funded research studies and developed several data analysis applications.

 
 

Time Estimation Tool Screenshot

Highway Contract Time Estimation Tool

Federal Highway Administration, Transportation Construction Management (TCM) Pooled Fund (TPF-5(260)

We developed an Artificial Neural Network (ANN) to predict the duration of highway construction projects from bid quantities and project characteristics. The ANN proved to be considerably more accurate than traditional linear regression models developed by other institutions, so we packaged the ANN into a web application that is deployed across the state of Colorado for use on Colorado Department of Transportation projects.

https://pmodev.codot.gov/contracttimes


Visual Eyes

We’re developing an app that provides new ways of analyzing and visualizing eye tracking data and our objective is to determine how eye tracking data can be used to predict performance. Specifically, we are currently working with two eye tracking datasets: a construction hazard identification dataset, in which we are predicting hazards identified; and a PVC pipe spool assembly dataset, in which we are predicting assembly time, % rework, and number of errors made.

https://github.com/mattsears18/visual-eyes

Visual Eyes Screenshot


Proposed Research Roadmap

How to Double Productivity

Construction Industry Institute, Research Team RT-DCC-01

“RT-DCC-01 was charged by the CII Downstream and Chemical Sector Committee to develop a research roadmap that if implemented would develop the knowledge and practices to double productivity of the industrial downstream and chemical construction sector.”

https://staging.construction-institute.org/resources/knowledgebase/knowledge-areas/construction-execution/topics/rt-dcc-01


Express Delphi

Construction Industry Institute, Research Team RT-DCC-01

As a component of the CII How to Double Productivity study, we developed a web application for ranking alternatives through the Delphi technique. Traditionally, the Delphi technique requires several weeks or even several months of data collection via email or paper forms and a considerable effort to assess the results. We used our Express Delphi app to conduct our study in 1–1/2 days, completely electronically, with all results calculated in real-time. Our participants were physically present in a room for discussions, but they could have easily worked remotely via video conference.

https://github.com/mattsears18/express-delphi

Express Delphi Walkthrough (No audio)


Example of a convex hull visualization produced by the Visual Eyes app

Time Hulls

While developing the Visual Eyes eye tracking data analysis app, it became apparent that the convex hull analysis method was applicable to datasets other than just eye tracking data. In fact, the convex hull analysis method can be applied to any timestamped 2D point data. We decoupled the code responsible for computing the convex hulls and released it as a stand-alone JavaScript package and the Visual Eyes app serves as an example implementation.

https://www.npmjs.com/package/time-hulls