Categories


Authors

Darkhorse Analytics

Darkhorse Analytics

Making the leap between gathering data and solving your company’s biggest challenge is a large one. Darkhorse Analytics, founded in 2008, makes that leap day-in-and-day-out for their clients; helping them solve complex problems by making data make sense.

From helping parents find high-performing high schools based on high school grade data made available through the provincial open data catalogue to making the case for roster changes for the Edmonton Oilers, Darkhorse Analytics has found success by focusing on three simple tenets: answer a single question, find metrics that matter, and share the solution in a clear visual manner.

We sat down with co-founder Daniel Haight at Transcend Coffee to chat data, visualizations and building a company in Edmonton.

How does Darkhorse Analytics help solve complex problems?

We help solve our clients’ challenges through analytics and data visualization. Analytics is applying math to a set of data to evaluate what’s happening, predict what’s going to happen, and prescribe what you should do about it.

Data visualization is about helping with adoption of that solution. Having the right answer is only half the equation, convincing a decision maker to act on that answer is much harder. We do that through data visualization — presenting the solution in a clear visual manner. 

Why did you start the company?

Joey Cherdarchuk, my co-founder, and I want to do meaningful work because it’s a great predictor of personal happiness and success. We don’t want to do a report that sits on the shelf; we want to do something that moves the dial.

We were working together at the Centre for Excellence in Operations at the University of Alberta. In the research setting, there would be pressure to find projects that would fulfill three areas: secure outside funding, result in academics publishing papers, and help organizations solve problems. Pushing the research as far as we could and solving real-world problems was our sweet spot.

We really hit a home-run in emergency services optimization. In Calgary we were able to crack the deployment nut. We could predict the impact on response time of adding or moving vehicles, and we could optimize these decisions to make the system perform better. We had a bunch of people knocking on our doors asking, “What you did in Calgary, can you do here?” In that short amount of time, we had three PhD theses published and worked for 15 municipalities.

It came to a point where we couldn’t push the research any further, but there was definitely benefit to the organizations. One former ambulance chief said “You should go into business. You should just do this.” So, we used all of our work for emergency response organizations as the launching point for Darkhorse.

It’s one thing to start a company to fill a market need, but since that time, a really big part of our journey has been asking, “how do we craft this company?” And, surprisingly, that’s what I’ve found I love the most. I love analytics and visualization, but I really love building a company.

How did you assemble your team?

We assembled through fits and starts. Currently we have 13 staff, including interns. Half the staff is on our development team (computer science and engineering) and the other half are more analytical types, coming from math or operations research.

If you draw four circles — data, code, math and visual design — the types of people we look for fall in at least two circles, preferably three. The intersections between those skill sets is where lots of interesting work happens. Joey is a great example of that intersection. He has developed depth in visualization, but has breadth across all the other areas. His work is leading the field and inspiring our team to aim for mastery.

Why build Darkhorse in Edmonton?

I grew up here; I have family here. Over time, I’ve come to appreciate that we have very skilled people coming through our schools and we can sell good, quality people on our unique work culture. We are fighting with Google and Facebook for some of the computing science and engineering grads, and we compete for applied math students with the oil and gas industry, but a lot of these talented young people want to try something different.

We play hard on weekends but when the sun goes down at 4:00 p.m. there are still people working. There is a strong work ethic, a loyalty, and high productivity with people who work here.

What was your most rewarding and challenging project so far?

We’ve hit some very challenging mathematical problems with our clients. For example, log flow analysis for a forestry company.  Log flow is how our client harvests, transports, and processes trees into pulp across the northern half of the province.

In the past, they had a manual method to solve the problem. Throw people in a room for a week at a time to manually solve it — six white boards and don’t come out until you have an answer. “When do we cut the tree?” “What trucks do we assign to it and when?” All these crazy constraints. “We can only truck out of these areas in the winter, but only in the winter when there is snow. Or else we have to truck snow up from a hockey arena to build ice bridges to safely cross the creek beds.” “This crew is predominantly Russian Orthodox, so we can’t operate on these dates…” Constraints, that change site to site, year to year. Very intense constraints. This was in 2011, and it was the biggest project we had taken on to that point. We knew theoretically how solve it but when you get into the actual process, and you see all the complexity, it’s “Oh my goodness, are we actually going to pull this off?”

We ran into the limitations with both the technology and the software. We had to come up with some very unique algorithmic solutions make it work. It took us an extra year to complete the project. They were very happy in the end; we solved their problem. It gives you a different appreciation for lumber.

On our forecasting projects, we’ve done a number of forecasting projects to predict future sales, costs, budgets or populations — where we will build a forecast model and it might take three or six months. In that time the client is collecting new data, and that we’ve not seen. In this final test meeting, we tell the client “we’re going to tell you what we think happened in the last six months and you can tell us how accurate we were.” It’s a true forecast.  We call it THE BIG REVEAL. Those meetings are the most rewarding because they’re like, “you’re way more accurate than we were!” Having the discipline to test your predictions builds a lot of credibility with clients.

Darkhorse Analytics uses data to identify new opportunities, algorithms to increase efficiency, and visualizations to aid decision making.

Interview by Stephanie Enders. Illustration by Amanda Schutz.

This interview was originally published in RISE, a publication about the risk takers who are shaping Edmonton. You can learn more or buy copies on the RISE website.

The Earth Group

The Earth Group

Integration Pilates

Integration Pilates