For the first time this semester, a cohort of junior QUEST students took the capstone 490H class. Team HAAT Consulting, working for CFR Engineering, was the first team to win the newly titled Outstanding Capstone Award. QUESTPress sat down with the team to discuss their project, inspirations, and advice for future 490H students.
The team was composed of Tommy Johnson (Electrical Engineering), Halley Weitzman (Computer Science), Aaron Sirken (Mechanical Engineering), and Avi Silvermetz (Finance), all from Cohort 21.
CFR Engineering is a small, 30 employee engineering firm located in Germantown, Maryland. They mainly provide electrical, mechanical, and plumbing/fire protection engineering solutions and services to their clients. CFR realized that much of their data was scattered and disorganized, and as such tasked HAAT consulting with constructing a query-able database for their bid and proposal data. They hoped to better be able to identify past bid and proposal data to assist in creating new bids for current and future projects.
What was your team’s final solution?
Avi: Our team created an Access database for CFR engineering that allows them to easily and efficiently query past projects in their database and quickly get the information they need from the data. We also created a multiple linear regression model to help them predict their future bid and proposal costs, and we provided them with a framework for expanding upon our work and taking our recommendations to the next level using machine learning.
How did you come to settle on this final solution?
Aaron: For the longest time, we were hung up on the idea of creating a custom database for CFR. We wanted our solution to be unique and have a number of custom features that we thought would add significant value to our client. However, as we worked through the project, our clients kept stressing how they wanted the deliverable to be simple and easy to use. We were running into a number of roadblocks designing a custom database, particularly the fact that we didn’t believe it would be sustainable for CFR. We knew that we needed to find a better balance of our wants with the client’s needs. Eventually, Avi suggested we look at Microsoft Access.
Avi: We knew that CFR had a license to use Access and that some of their employees on the business development team had experience using it. So that already reduced a few of the roadblocks to creating the database. It also interfaced easily with Excel spreadsheets, and eventually, we realized that Access was what was best for our client.
How did you design the linear regression model?
Halley: We initially wanted to use machine learning to estimate project costs. However, we realized that was too ambitious of a goal for the scope of our project and time remaining in the semester. The linear regression model was our next attempt to help CFR predict their project costs.
One of the challenges in building the model was the vast amount of data. We had to eliminate a lot of it to make the model better and find the correct predictors of project cost. Dr. Armstrong, our faculty advisor, was a huge help in formulating the model, even meeting with us on weekends to lend her advice and assistance. One cool thing about the model is that it performs better on data it hasn’t seen before, which we believe will help it to accurately predict project costs for CFR.
How did you come up with your +1 or next steps?
Tommy: We were sitting in the QUEST lab a couple of weeks ago working on our poster and looking at past posters, and Dr. Bailey walked in the room. We began talking with him about our client and project, and he immediately understood our project better than even we did. He started writing on the white board and walking us through every aspect of our project and the work we did, really helping to identify what we needed to put on our poster and how to talk about our project.
He then told us that 490H projects are 80% scoped out and 20% up to the team to have creative flexibility with. He said that while we had achieved the 80%, this was our opportunity to add value to CFR with our 20% recommendations. He walked us through how our work could be expanded upon and improved, using our original idea of machine learning, to turn a 10-million dollar business into a 10-billion dollar business. He was the inspiration behind our final recommendations, and he really helped us reexamine our project and scope and break out of a bit of a lull we were in at the time.
What advice would you give future 490H students?
Tommy: Deliver what the client wants, but don’t be afraid to take advantage of the 20%. It is important to always keep the clients’ needs in mind and build for them, but this is also a learning opportunity and your chance to leave an impact on the real world, and that’s where the 20% comes in.
Avi: Although this project was a great learning experience and huge success, I still wish I had pushed my boundaries a little bit more and tried something new. I did a lot of things for this project I already had experience in, and I wish I had tested new skills during my 490H project.
Aaron: Our team got along really well. Our meetings and discussions were all really lighthearted and fun, and we didn’t get frustrated or argue. I would encourage teams to try to keep things as lighthearted as possible while still getting your work done.
Halley: Focus on the client and the solutions the client wants. We lost a lot of time trying to build solutions they couldn’t use. Start EARLY. Front load as much of the work as possible. And don’t be afraid to distribute and delegate work. Not everyone needs to be present to do everything, and you will accomplish tasks more efficiently by trusting your teammates and delegating tasks.
