Women in Optimization: Dr. Vanesa Guerrero

This article is the fourth in our series, “Women in Optimization.” Although these articles will highlight women, we hope the stories will encourage all individuals who might want to pursue a career in optimization.

As a researcher and mathematician, Dr. Vanesa Guerrero has led an inspiring career working at the intersection of statistics and mathematical optimization. Now, as a professor at the Universidad Carlos III de Madrid, she guides new generations of students through the dynamic relationship between these two fields.

From navigating different terminologies to completing the challenging road to professorship, Dr. Guerrero’s work has helped to bridge the gap between statistics and optimization while shaping the path for future professionals.

A Tale of Two Disciplines 

During her undergraduate studies, Dr. Guerrero’s statistics professor, Dr. Emilio Carrizosa, encouraged her to apply for a research training scholarship. She won the scholarship, and Professor Carrizosa would later serve as her PhD supervisor.

“At that point, both Professor Carrizosa and my other supervisor, Professor Dolores Romero Morales, had already published many works on the intersection of statistics and optimization, and how both disciplines can help each other. So of course I fell in love with the topic, and it continues to be the main focus of my research,” says Dr. Guerrero.

While there are indeed many ways that the two disciplines complement each other, there are also some unique distinctions that can make working between the two a challenge.

“There are some important differences in terminology,” explains Dr. Guerrero. “For example, statisticians refer to variables as inputs that are known, while parameters must be estimated. But for operations researchers, it is actually the opposite—parameters are the given input, variables are unknown, and you need to determine the optimal value of those variables.”

So how does one overcome such extreme differences?

“I think that by just being patient and adopting the vocabulary, you’ll end up succeeding,” she says. “But in general, in any interdisciplinary project, it’s very important that the vocabulary is well established, and that you convey your message with care. This is of course a challenge, but I’ve learned a lot about how to communicate and deliver my message in a way that will be easily understood by the target audience.”

The Path to Professorship

Despite conquering the difficulties that come from working at the intersection of two distinct disciplines, Dr. Guerrero wasn’t always confident in her path to professorship. After completing her PhD, she was unsure if she would be able to achieve tenure within the eight-year timeframe specified by her new contract.

“After I finished my PhD at the Universidad de Sevilla in 2017, I started applying for postdoc positions and also tenured associate professorships, without a lot of hope,” she recalls. “But I received a call from Universidad Carlos III de Madrid offering me a tenure track associate professorship position, and in the end I decided to take it, even though I thought I would not be able to get tenure in eight years, since my list of papers wasn’t very long.”

Dr. Guerrero persisted with the same strategy that has helped her through her life and career: taking things one step at a time.

“The first year was hard, both professionally and personally. The working environment was different from the one I was used to, I was in a new city, and I had to make new friends,” she says. “But I was lucky, because I had the opportunity to start collaborating with people from different departments, and I feel very grateful for them because they believed in me even when I didn’t.”

In the fifth year of her associate professorship, Dr. Guerrero obtained what is known in Spain as the acreditación by the National Agency for Quality Assessment and Accreditation (ANECA), a certification that allows one to apply for associate professorship positions. She successfully achieved tenure and is now an associate professor.

“When you have these specific goals that you need to reach—publish a certain number of papers, attend a certain number of conferences—I think it’s important to say, ‘Okay—one thing each day’—because there are things that you cannot really control, like the time it will take for the paper to get published,” she explains. “So you keep moving forward, and things will come at the moment they’re meant to.”

Combining Statistics with Optimization

Dr. Guerrero’s current research is still at the heart of both fields—statistics and mathematical optimization. How does mathematical optimization support statistics?

One example can be found in the feature selection problem, which involves selecting a set of features (also called “variables” or “predictors”) that explain a certain outcome or target. These problems originally stem from the field of statistics, which Dr. Guerrero approaches from a mathematical optimization perspective.

Say you are implementing a system to identify spam email messages. Some explanatory features may include the frequency of certain words in an email, the time of day the email was sent, or the domain of the sender’s address. The feature selection problem can help you identify the best features for determining whether an email is spam or not.

Another example from the field of medicine is biomarker discovery. Imagine that a doctor needs the value of a certain biomarker to make a diagnosis, but obtaining this biomarker is difficult and requires an invasive procedure. Biomarker discovery involves estimating the value of the biomarker from a set of other variables that are easier to obtain, such as blood test results, height, weight, age, or genetic information.

The feature selection problem would answer questions such as, “Which features are good at predicting the outcome?” If the number of features is very high, you could also ask, “How can I summarize this information in a smaller set of features that can still retain the output information of the larger set of features?” By leveraging the flexibility of mathematical optimization problems, you can create a framework where doctors can easily add or remove features or explanatory variables to the problem.

This relationship works both ways: not only can mathematical optimization support statistics, but statistics can also enhance mathematical optimization.

For example, Dr. Guerrero has worked with huge mathematical optimization problems that are very difficult to solve from a computational perspective. She leverages statistics to build approximations of the constraints or the objective function of those problems, so that they become computationally tractable while remaining as accurate as possible.

Inspiring the Next Generation of Women in STEM

While Dr. Guerrero loves being involved in a variety of different projects, she’s learning to say ‘no’ more often, in order to dedicate more time to the work she’s truly passionate about. One of the projects she’s dedicated herself to outside of her research work is a mentorship program, the Data Science Gymkhana, which introduces young students to the importance of data analysis in different contexts, such as sustainability initiatives.

“It’s a really nice initiative that promotes the STEM (Science, Technology, Engineering and Mathematics) disciplines for the younger generations,” she explains.

One of the most rewarding parts of her work is the opportunity to collaborate with so many inspiring professionals.

“There are so many opportunities to meet, work, and collaborate with different people—you can really decide who you want to work with, and that’s something not many people are able to do,” she shares. “There’s also a lot of flexibility around the topics you can research, as well as your schedule—which I think is essential for maintaining balance.”

For women in particular, having strong role models and representation in STEM is essential—but as Dr. Guerrero notes, there is still work to be done.

“We have some great examples of female leadership in our community, and of course progress has been made. But there are still some structural barriers,” she says. “I think especially when it comes to balancing academic life with family life, things become typically more challenging for women than for men. So we need to help each other out and be thoughtful about how we mentor and collaborate with one another.”

Finally, she advises all aspiring STEM professionals to enjoy the journey, and try to learn as much as they can from their colleagues.

“If I could say something to my younger self, I would say take it easy. Enjoy the process, work hard, and try to learn something from all the amazing people that you are going to work with along the way,” she says. “I think that is truly one of the most valuable opportunities we have in this field.”