Skip to main content Skip to footer

FORO

Machine Learning Application Saves an Organization Over $100 Million

Using a machine learning algorithm and custom user-interface, Parux helped FORO and the Indiana Department of Transportation develop a solution to bundle road-construction projects together to save money. The savings were massive which will allow INDOT’s budget to go further and improve operational efficiency.

The What –
A Cloud Platform to Bundle DOT Projects

FORO works with Parux to design and develop a cloud platform that leverages uses machine learning to help support their client bundle road, bridge, and other transportation projects.

The Why –
Improve Efficiency of DOT Budgets

Bundling of construction projects is a time-consuming process that is very limited. Using a machine learning model, bundles can be created faster with more efficient results.

Construction site with a small CAT loader, pipes, and materials laid out near a park.

The Brief –
FORO Utilizes Parux to Help Them Improve Bundling for INDOT

FORO is a growing start-up that promises “Better Insights. Better Decisions.” Its core offerings are robust applications and consulting services for large companies and organizations that need complex data processing, verbatim text analysis, and improved team collaboration. Parux has worked with FORO for years to develop its standalone SaaS product as well as custom applications for FORO’s clients.

The Indiana Department of Transportation (INDOT) manages hundreds of state construction projects every year, and is a leading state in implementing a cost-efficiency method of project bundling. By taking several small projects, and attaching them to a large project, the cost savings are significant for the State of Indiana. FORO brought in Parux to design a bundling application to help INDOT more efficiency created their bundle program.

  • Ideation
  • UX Design
  • UI Prototyping
  • Machine Learning
  • Development

The Senior Leader –
Meet Greg

Greg is the Director of Asset Management at a state Department of Transportation. He is responsible for planning and forecasting all of the state’s transportation construction projects. He’s an advocate for bundling projects together to help maximize the state’s budget.

Greg, persona representing the senior leader

Personality

  • Tech Savvy 4/10
  • Analytics Centric 9/10
  • Open to Innovation 8.5/10
  • Organization 8/10

Pain Points

  • Bundling thousands of projects together takes weeks to complete for 3-5 people.
  • Has to set restrictive parameters when manually bundling projects.
  • Unknown whether manual process is leading to optimized results.

The Problem –
How Can FORO Improve Project Bundling for INDOT?

INDOT for years has evaluated their project list and gone through a process to bundle like projects together. This process takes proximity of the projects’ mid-points together as well as other factors like type of work, budget size, and union contractor affiliations.

  • Manual process takes up significant INDOT resources for weeks.

  • Unknown whether manual process was leading to optimized results.

  • Bundling based on a mile-radius of project mid-points is an efficiency limiter. Some projects span dozens of miles while others are at a single intersection. However, bundling by radius based on start and end points is too complicated to do manually.

  • INDOT has to set arbitrary rules to confine the scope of bundling such as not projects not crossing union and non-union lines.

The Solution –
Introducing FORO DDA - A Data and Decision Analysis Solution

FORO and Parux developed a custom web-base application to be utilized by the Indiana Department of Transportation. The primary goal was to give INDOT the ability to create a full-set of bundles quickly, but keeping the human-element present to properly leverage the indicate knowledge of the INDOT staff. The core of the solution centered around a custom machine learning algorithm that takes the tedious manual bundling process and streamlines it with just a few clicks. The machine learning base was wrapped in a simple-to-use interface that allows INDOT to make adjustments to the core data set and project bundles.

Dashboard interface for project bundling statistics. An activity log is shown on the right.

Designing the Platform –
Connecting the Dots on Construction Projects Using Machine Learning

The first was for FORO and INDOT to work with Parux’s AI/Machine Learning team to develop the rules and criteria needed to create project bundles. Parux developers took the criteria given by INDOT and worked on creating the machine learning algorithm which was facilitated through a custom-built API. The algorithm went through several iterations to increase the cost-efficiency of the produced bundles. Each iteration of the machine learning algorithm was compared against a bundle set manually created by INDOT. The end product was a proof-of-concept application that demonstrated the results of the various runs types that were created and refined that produced the most efficient bundles.

Abstract network visualization with connected nodes, displaying various numeric values.

Designing the Platform –
Building a Full-Service Interface to Manage Bundles

Once the bundling methods were established, an online application was designed and built for the INDOT staff to run the projects through the machine learning bundling application. The Parux UX/UI team held several discovery sessions to discuss the workflow process used to create Programs (the set of bundled projects). Prototypes were designed and reviewed by FORO and INDOT, and then were developed into a working application. The final interface allows INDOT staff members to review, edit, and select the best produced bundles to complete its Program for the next fiscal year.

Dashboard interface displaying details for multiple project bundles.
All Data is Randomly Generated

Making an Impact –
The DDA Platform Has a Huge Impact

Extra $107,939,462 in Cost Savings

The machine learning algorithm projected to save INDOT an additional 40% in project cost. These extra cost savings allow INDOT to increase the number of projects in a fiscal year and take on more large-scale specialty projects.

Bar chart comparing savings between FORO process and manual process from 2021 to 2024.

From Weeks to Days

The application maximizes efficiency with customized decision-making. INDOT will now be able to develop a Program in just a few days, instead of the weeks the manual process took.

Calendar showing days marked for manual process in orange and FORO process in purple.

Better Bundles

The FORO results were also more accurate. With the manual process, assumptions had to be made on bundling to get the Program completed at all. FORO challenged those assumptions and produced a more accurate bundle set with new scenarios that produce better results.

Simple map illustration with orange location pins along intersecting lines.

The Result –
INDOT Uses the FORO DDA to Save Hundreds of Millions Every Year

The Data and Decision Analysis application has proven to be a huge success for FORO and INDOT. The machine learning algorithm has produced more accurate Programs while saving INDOT weeks of time - making them more efficient. FORO is now in the process of on-boarding more states to use its system.

Key Result

The FORO DDA Platform has gone from a machine learning proof-of-concept to a full-service application that INDOT actively uses in their yearly bundling process. The machine learning algorithm has now been extend to include new rating factors such as equity and resiliency of each project.

The process to bundle projects together is vital to INDOT. However, the pain-points were very real. It involved several staff members meeting for weeks trying to find the right Program to implement. INDOT knew they were saving money but had to no idea if they were maximizing those savings. They knew they needed our help. Parux took us from a concept to a full MVP. Our platform’s flexibility has allowed us to grow and transition seamlessly.

Brett Boston Founder & CEO of FORO
Brett Boston portrait
Download FORO Case Study as PDF
Read next Case Study: InScope