Author Archives: Laura Albert

COVID-19 is a pandemic that requires systems thinking and solutions

I was on the INFORMS Resoundingly Human to talk about COVID-19 and first responders. You can listen here:

https://pubsonline.informs.org/do/10.1287/orms.2020.02.11p/full/

In the podcast, I discuss supply chains, rationing resources, and disaster planning, and I note how everything old becomes new again. For example, the US is not experiencing its first N95 mask shortage. Systems concepts are important for understanding how to prepare for and respond to a pandemic.

In this post, I want to dig deeper into systems concepts. I wrote a quick primer on systems thinking and explain why systems concepts are important for understanding the COVID-19 pandemic.

What is a system?

A system is a set of things—people, vehicles, basketball teams, hospital beds, or whatever—interconnected in such a way that they produce their own pattern of behavior over time.

Here are three examples:

(1) A car is just a vehicle. A collection of cars can be a traffic jam.

(2) A single ventilator can be used to treat a patient. A hospital’s collection of beds and ventilators are available for treating patients. When a surge of patients require these resources, they may have to wait and queue for these limited resources.

(3) An N95 mask protects first responders from infectious disease when they treat patients. A supply chain of personal protective equipment (PPE) can have delays and shortages, leading to first responders not having the N95 masks they need at any given moment.

How is systems engineering relevant to COVID-19?

COVID-19 is absolutely a medical challenge. It is also a systems challenge that require systems thinking and systems solutions. In systems, decisions are not made in isolation, but rather, decisions are interrelated.

My discipline is operations research: the science of making decisions using advanced analytical methods. Systems require a series of decisions to operate effectively with or without patient surges in a pandemic. Operations research provides the analytical tools required to design and operate systems more effectively and efficiently.

In systems there are many trade-offs and complicated interactions. Here are examples of how systems engineering is important now:

(1) If a first responder does not have adequate personal protective equipment (PPE) such as latex gloves and N95 masks, they are at higher risk from acquiring COVID-19. If they do, they will not be able to treat patients in the coming months, thereby reducing the number of first responders (a critical resource) in the future. This informs how responders should treat patients and ration resources now.

(2) Surges in COVID-19 cases may lead to more patients requiring ventilators than are available in hospitals. This could lead to rationing and painful choices that would not be considered without a patient surge.

Systems concepts will continue to be important in the future. Here is a third example:

(3) One person who gets a vaccine has immunity. If enough people receive vaccines or have immunity from previously having had the disease, we can achieve herd immunity and eliminate person-to-person transmission of the disease even among those who do not have immunity. With herd immunity, the benefits are greater than the sum of its parts.

What can systems thinking tell us about the fatality rate for COVID-19?

It depends. We know that it depends on age, gender, and co-morbidities. The fatality rate is not an exogenously given number, but rather it is a function of the resources available for treating patients, which is endogenous to the system. The fatality rate for COVID-19 is a systems concept. If the number of infected individuals is low enough so that hospitals can handle the surge and give every patient the treatment they require, the fatality rate will be lower (relatively speaking. In absolute terms it will still be too high). The fatality rate will be a lot higher if hospitals are over capacity and have to ration beds and ventilators.

How are my personal decisions related to healthcare systems in the COVID-19 pandemic?

The resources in our healthcare system are being stretched to the limit. The resources include personnel (physicians, nurses, first responders), hospital beds, ventilators, and personal protective equipment. When there are not enough resources to give every COVID-19 patient the best treatment they require, physicians will have to ration resources and make tough choices. Our efforts to delay the second wave as long as possible and to reduce the number of people who require medical treatment will save lives. Flattening the curve is a systems concept aimed at reducing painful tradeoffs and complicated interactions.

How can we prevent the next wave?

Preventing the next wave of any infectious disease is a numbers game. I do not know how to practice medicine but I know how to crunch numbers. The key is to lower the overall transmission rate. The best way to lower the transmission rate varies according to the disease, but there are some basic principles for preventing a disease outbreak from becoming another wave of a pandemic. Best practices include better hygiene practices such as washing your hands and your mobile phones with soap and water, and covering your cough. Limiting the number of people you come in contact with reduces the opportunities for transmission. All those trips to the store to buy extra toilet paper increase one’s chance of contracting COVID-19.

What can we do to prepare for a second wave?

A second wave in a prolonged pandemic is not going to be easy for many of us. I use mathematical models and analytics in my research, and I find them to be useful in my everyday life. My research tells me that I make better decisions with better information and that I should use limited resources wisely. When I think about what it means to apply these principles to my decisions in a pandemic, I realized I can achieve both of these goals by gathering up to date information and following instructions from official, trusted sources such as local and state governments, local police and emergency medical service departments, and the Centers for Disease Control and Prevention. I plan to use the official sources to limit what I think about, worry about, and do in any upcoming waves of the pandemic. We are all inundated with conflicting information and advice from many sources, and it is taking its toll and potentially leading us to make unsafe choices such as making repeated trips to grocery stores to stockpile items we do not need.

 

Related posts:


Travel Bans Can’t Stop this Pandemic

My op-ed entitled “Travel bans can’t stop this pandemic” was published in The Hill. You can read it here.

 


emergency response during mass casualty incidents

Today’s blog post about my research on mass casualty events and emergency response given that COVID-19 has been declared a pandemic by the World Health Organization (WHO). I have four papers in the area that are relevant in the area of emergency medical services (EMS) during mass casualty incidents.

A mass casualty incident (MCI) is an event in which the demand for service overwhelms local resources. Since fire and EMS departments operate at the local level, they can be overwhelmed quite easily. Anything from a multiple vehicle accident to a weather disaster to a hospital evacuation can be considered an MCI. Fire and EMS departments have “mutual aid” agreements with neighboring departments to address the more routine of these incidents and have “Standard Operating Procedures” for a range of more severe incidents. However, switching between such policies in practice is not simple. Moreover, not all mass casualty incidents are the same. Responding to calls for service during a hurricane is different than during a pandemic. In the latter, paramedics and emergency medical technicians can become sick and should stop treating patients, leading to fewer resources for responding to patients that require service. Additionally, we would expect less road congestion and wind during pandemics than in a hurricane evacuation. However, both cases may see a surge of low-acuity patients who request service.

My research focuses on emergency response during MCIs lifts limiting assumptions made by papers in the literature, which often assume that there are enough resources available all the time (which is not a reasonable assumption during MCIs). Here is a summary of four of my papers that have addressed MCIs.

Dubois, E. Albert, L.A. 2020. Dispatching Policies During Prolonged Mass Casualty Incidents. Technical report, University of Wisconsin-Madison.

The newest paper is available as a technical report and is most relevant to COVID-19. It focuses on a large surge of patients that overwhelmed EMS resources. Here, we lift the assumption that a patient’s priority is a fixed input. Instead, we consider patients whose conditions deteriorate over time as they wait for service.  We consider how to assign two types of ambulances to patients, advanced and basic life support. We study how to dispatch ambulances during MCIs while allowing ambulances to idle while less emergent patients are queued. This is similar to keeping a reserve stock of advanced life support ambulances (see the last paper listed in this post). The inherent trade-off is that when low-priority patients are asked to wait for service, they can become high-priority patients. When high-priority patients are asked to wait for service, they can become critical or die. Our solution method is to find dynamic response policies to match two types of ambulances with these three types of patients.  We observe that, under the optimal policies, advanced life support ambulances often remain idle when less emergent patients are queued to provide quicker service to future more emergent patients. It is counter-intuitive to not use all resources all the time during an MCI. However, keeping some resources in reserve ensures that there are resources available at the time the most critical patients need them.

McLay, L.A., Brooks, J.P., Boone, E.L., 2012. Analyzing the Volume and Nature of Emergency Medical Calls during Severe Weather Events using Regression Methodologies. Socio-Economic Planning Sciences 46, 55 – 66.

The second paper seeks to characterize the volume and characteristics of EMS and fire calls for service. It was motivated by the need to deliver routine emergency service during weather emergencies and disasters. What typically happens during emergencies is that there are more calls for service, most of which are low priority calls. Triage becomes more important in these situations, because the most severe calls for service can be drowned out by so many low-priority requests. However, call surges are not the only stress on fire and EMS departments. Road congestion and slow travel times mean that each call takes more time to serve, which can further stress limited resources. As a result, it becomes important to triage calls and assign appropriate resources.

Kunkel, A., McLay, L.A. 2013. Determining minimum staffing levels during snowstorms using an integrated simulation, regression, and reliability model. Health Care Management Science 16(1), 14 – 26.

The third paper studies staffing levels during a blizzard, where a surge of calls can temporarily overwhelm resources that are available. Additional staff are usually scheduled during emergencies when call volumes increase. We specifically focus on snow events, and the results have insight into other situations. To determine staffing levels that depend on weather, we propose a data-driven model that uses a discrete event simulation of a reliability model to identify minimum staffing levels that provide timely patient care,with regression used to provide the input parameters. We consider different response options, including asking low priority patients to wait for service, and we take into account that service providers often work faster when systems are congested. The latter issue of allowing adaptive service rates is important, since it makes the model more realistic by limiting the assumption that service rates are constant. A key observation is that when it is snowing, intrinsic system adaptation with respect to service rates has similar effects on system reliability as having one additional ambulance.

Yoon, S., Albert, L. 2018. An Expected Coverage Model with a Cutoff Priority Queue. Health Care Management Science 21(4), 517 – 533.

The final paper examines how to locate and dispatch ambulances when resources can be temporarily overwhelmed. In this paper, there are prioritized calls for service in a congested system, but the system is not completely overwhelmed by an MCI such as a hospital evacuation. Typically, models in the literature implicitly assume that there are always enough resources to respond immediately to all calls for service that are received. This is not a good assumption when there is an MCI. As a result, we need new models and analyses to provide insights into how to allocate resources when there is congestion and many service providers are busy treating patients.

We formulate new models to characterize policies when ambulances are held in reserve for high priority calls. When the system is so congested that it hits the “reserve” stock of ambulances, low priority patients are either diverted to neighboring EMS systems through mutual aid or added to a queue and responded to when the congestion has reduced. Interestingly, we find that by adopting such an approach for sending (and not sending) ambulances to patients, this affects where we might want to locate ambulances at stations.

 

 

 


Angst: Preparing for an overseas sabbatical

Preparing for a six month sabbatical overseas was daunting. My sabbatical takes me to Aachen, Germany from January until July, a span of six months. In addition, I had to prepare for my daughters for this trip. This blog post is about how I moved with my family to a new country, focusing on month leading up to the sabbatical once I knew where and when I was going. A post about the longer-term planning will come arrive.

Paperwork

Thanks to Sara Zaske (author of Achtung Baby, which I highly recommend), I knew I should print out and prepare folders full of important copies of documents to navigate German bureaucracy. This was the smartest way I prepared. Some of the documents I packed include various forms and instructions from the Fulbright Commission.

  • Copies of Fulbright documents, including the letter indicating I could work in Germany for longer than 90 days
  • Copies of the invitation letter from RWTH Aachen
  • Copies of health insurance (required for living in Germany)
  • Birth certificates (with extra copies)
  • Copies of passports
  • Copies of report cards (for my daughters)
  • Copies of all travel documents

Deciding what to pack

It became clear that I would need a lot of outfits for work, conferences, relaxation, and running. I also needed shoes for all these activities. I used the Set Cover problem to decide what clothes and shoes pack that could “cover” most of my outfits across these activities I would take part in. It became immediately clear that my shoes and wardrobe would have to be compatible with the color black if I wanted to fit my clothing into one large suitcase. I had to pack shoes for many occasions, and I decided that my outfits should mostly go with a series black shoes (sandals, flats, pumps, booties, boots). Gray was a second choice for color compatibility. I left a pair of jeans at home because they did not go with enough outfits or shoe choices. My kids packed a few small games for us to play in Germany. It was nice to have some entertainment.

Preparing my devices (phone and iPad) for the sabbatical was one of the most important ways I prepared for the trip.

  • I downloaded a map of Aachen on google maps.
  • I downloaded the google translate app on all of my devices, and I downloaded German so I could translate when offline.
  • I downloaded the latest version of google voice on my phone so I would have access to a stable US phone number that could make phone calls and receive SMS texts.
  • I downloaded a few books on the Kindle for my children to read.
  • I created a WhatsApp account that I started using with folks in Germany.
  • Other apps included Deutsche Bahn, a conversion app (for distances, temperatures, etc.), and Aachen’s public transit app.
  • I forwarded all of my travel confirmations to Trip It.

I purchased and/or a few items that were extremely useful before my trip:

  • Outlet adapters with USB adapters (Six was about right for the four of us). We like the ones with USB ports.
  • I bought a discount Deutsche Bahn card (the Bahn 25) for discount rail tickets. I purchased a flexible, discount ticket for my trip from the Frankfurt airport to Aachen ahead of time.
  • I bought a new Amazon Kindle Fire for my kids to read books abroad, since English speaking books would be hard to find. We checked out copies of digital books from our library in Wisconsin, and we bought a few e-books as needed.
  • I bought a packable water-resistant, wind-proof Primaloft jacket from Lands End for a prior trip that was perfect for my sabbatical as an all-purpose coat in Aachen, where it has been cold, rainy, and windy.
  • A small wireless bluetooth speaker. It was nice to carry around a speaker and helped me stream podcasts and audiobooks while at home. But I could have lived without this.
  • A copy of Rick Steves Germany. I could not live without this.
  • Samsonite compression packing bags. I basically zip-locked my belongings in these large bags and squeezed the air out to pack more of my belongings into my suitcase more tightly. This was helpful for sweaters and other bulky items.

I packed a few extra tote bags and backpacks that could be laid flat in suitcases. These ended up being very useful, since we did not have a car and had to walk home while carrying our heavy grocery bags. I decided to purchase lotion, hair products, hand sanitizer, and other heavy toiletries in Germany to keep my bag lighter. This was a good decision, mostly because it was fun to shop for new items in Germany. I packed a few travel size shampoo and conditioner to tide me over. Toiletries and cleaning products are significantly cheaper in Germany. I also packed a few old towels, hot pads, sheets, fleece throws, and pillowcases to be able to use beds in the new apartment (and not make the return journey). In retrospect, I should have done some research on where to purchase thrifty towels and sheets near my flat. But it was nice having some immediate towels and hot pads to use. There are many discount stores in Germany, so it was extremely easy for us to buy what we needed in Aachen with very little effort. We bought all sorts of things at Aldi, including fleece sock liners, hoodies, sneakers (we quickly wore out the sneakers we brought due to so much walking), and even a coat when my daughter’s zipper broke. We are lucky that Germans like a bargain. There were several discount stores within half a mile of our flat.

COVID-19 broke out in Germany before I was able to stock up on disinfecting wipes and gel. That was unfortunate, but that was a fluke event that I could not anticipate. We are still able to wash our hands and use cleaning products. I will likely write a blog post just about COVID-19 at some point.

I decided to get a SIM card for my cell phone in Germany. My research indicated that Aldi Talk was the best plan in the country, but the registration process was difficult. I had to register for the SIM cards in Aldi with my passport, and the set up process took was confusing for me. It took me a few days to set up the SIM card, mostly because the PUK code did not work the first time I tried it, and I was confused on how to get my phone to recognize the new SIM card. It was a long time without a cell phone.

Preparing for life overseas.

I had to prepare to live overseas. I made a list and I started to think about changes months ahead of time. One of the challenges was to ensure that I could access all of my accounts that require two factor authentication (my university, google, dropbox, Paypal, Box, Amazon, others). I did not yet have a cell phone in Germany, so I changed the mobile phone number associated with my accounts to my google voice number. I signed up for as many email statements instead of mail statements for my bank, credit cards, insurance, and utilities. I unsubscribed from email lists for shopping in the US. I let my bank know I would be traveling. I made sure my credit cards and insurance accounts were set to auto-paid.

I suspended my car insurance for six months. I called to cancel various activities that my daughters were involved in (Irish dance, after school care, gymnastics).

Preparing my home for the move was easy, since my partner could not spend the entire sabbatical in Germany. He was able to look after the house and the cats aside from his planned trips to Germany to see us.

Preparing my children for overseas.

I had a plan for the children to take online classes while we were away. I was in contact with the children’s school teachers the year prior to moving. Fall parent teacher conferences in November was a good time to have a personal conversation with all teachers and sketch out a plan for the January transition. This made it easy to touch base to discuss follow up issues with teachers right before we left. I planned to fly to Germany the week after my oldest took her high school finals in January. This allowed the children to have a few days to transition to online courses before we moved. The children each brought a chromebook for school, headphones, mechanical pencils and pens to Germany.

I set up an appointment with folks in Germany (through the university and the cultural integration office) to find classes and extracurricular activities for the children. I found it helpful to be in touch with everyone involved and to be patient. Online school and life in Germany were different, and as a result, I was not always sure what questions I should ask. It took me awhile to wrap my head around what routines and schedules would look like.

Making the trip.

My children and I each packed one big suitcase and one big carry on.

They liked having a say in what they packed. A packing list helped, since that provided the guidelines for them. I packed a second carry on bag for extra items that was only half full. This way, we could manage our luggage from Madison to Aachen by plane and train. My daughters deviated from the packing list and added a few extra dense items (games, books, and lots of liquid toiletries). Two of our bags were overweight by a few pounds. My set cover approach was thwarted! But it worked out. I was able to tuck a few items into the half full carry on. When we picked up our luggage in Frankfurt, everyone was able to manage their own bags on the trains, and we made it to our new home in one piece.

Our new apartment was not fully furnished. I had to come up with a plan to furnish it ahead of time. I placed an order for IKEA furniture a week before I left and chose a delivery time for the day after I arrived. IKEA in Germany does not have next day delivery, and orders are delivered about a week after they are placed. I was incredibly nervous about my plan, but it worked perfectly. It was thrilling to get the delivery. We could get settled into our new home. The best part was that when my kids put together furniture and arranged their rooms, it helped them feel like it was their new home. I’ll write more about our first steps in Germany in another blog post.

In summary

The title of this post refers to Angst, because there was so much to worry about and so much in the air. This weight was most acute when it came to my concern for my three daughters. I felt the weight of this uncertainty and angst the most the month before leaving for Germany. I knew I would be find and could roll with the punches, but this trip was entirely new to them and was a stressful new experience. I was relieved when pieces of the planning fell into place.

In my next sabbatical post, I will blog about first steps in Germany. Read more blog posts about my 2020 sabbatical here.

 


STOR-I Masterclass at Lancaster University

Last week I traveled to Lancaster, England to teach a research masterclass at the Centre for Doctoral Training (CDT) in Statistics and Operational Research (OR) in partnership with Industry (STOR‐I) at Lancaster University. STOR‐i was established in 2010. It is funded by EPSRC, Lancaster University and a wide range of industrial partners. It’s goal is to use industrial challenges as catalysts for innovation, and the Centre’s primary aim is to develop future international research leaders in statistics and OR. A masterclass is a series of introductory talks on an area of contemporary research given to the PhD and Masters students enrolled in the program.

My masterclass was entitled “Public sector operational research.”

A brief description:

Public sector applications, such as those in fire and emergency medical services, are complex systems that span people, processes, vehicles, and critical infrastructure. Researchers have been developing optimization models to locate vehicles such as fire engines and ambulances and spatial queueing models for analyzing public safety vehicle deployment decisions for nearly 50 years. A body of literature for locating and dispatching vehicles has grown to lift simplifying assumptions and address important issues overlooked in the early research models in this area. Public sector applications such as homeland security, disaster preparedness and response, and critical infrastructure protection have received a growing amount of attention from operational researchers in recent years. However, many research challenges remain.

In this STOR-I masterclass, we will study the evolution of operational research in the public sector with application to public safety, homeland security, and disasters. Technical topics include network optimization problems, facility location and covering models; network design, restoration, and interdiction models; spatial queueing models; and discrete event simulation. Policy insights as well as issues relating to putting the results into practice in real-world settings in the United States and abroad will be discussed.

Readings I used in my lectures:

  1. Larson, R.C., 2002. Public sector operations research: A personal journey. Operations Research, 50(1), pp.135-145.
  2. Green, L.V. and Kolesar, P.J., 2004. Anniversary article: Improving emergency responsiveness with management science. Management Science, 50(8), pp.1001-1014.
  3. Reuter-Oppermann, M., van den Berg, P.L. and Vile, J.L., 2017. Logistics for emergency medical service systems. Health Systems, 6(3), pp.187-208.
  4. Albert McLay, L., 2015. Discrete optimization models for homeland security and disaster management. TutORials in Operations Research (pp. 111-132). INFORMS.
  5. Simpson, N.C. and Hancock, P.G., 2009. Fifty years of operational research and emergency response. Journal of the Operational Research Society, 60(sup1), pp.S126-S139.
  6. Ansari, S., McLay, L.A. and Mayorga, M.E., 2017. A maximum expected covering problem for district design. Transportation Science, 51(1), pp.376-390.

The masterclass was given in three two hour classes. While I was able to cover a lot of ground over six hours, I had to keep the scope relatively narrow so that we could discuss in depth. I decided to mainly focus on facility location models for siting resources for responding to routine and large-scale disasters.

Goals for the masterclass

Class 1: Public sector OR overview

Understand the history of public sector OR (in the US)
Evaluate when and how to apply public sector OR models
Understand features of emergency medical service systems and identify how these features can be represented in OR models

Class 2: facility location for emergency medical services

Understand and interpret facility location problem features
Apply facility location models to locating ambulances
Model how to locate ambulances by including increasing levels of model realism

Class 3: large-scale emergencies and disasters

Understand disasters concepts
Understand and interpret emergency management concepts for OR modeling
Apply OR models to disasters situations

I enjoyed getting to know faculty, lecturers and students. For example, I found out that there were three Slytherin in the class of about 40.

Six hours of teaching is a lot of teaching, and I’m grateful for the students who gave me their undivided attention for so long. One student had studied at RWTH Aachen (my host institution in Germany) and gave me a list of recommended things to do.

Read more blog posts about my 2020 sabbatical here.


firsts in operations research

The University of Wisconsin-Madison recently celebrated the 150th anniversary of the first women receiving college degrees at UW-Madison. It took almost 50 years after that for the first black woman to earn a UW-Madison degree. I am currently on my sabbatical at RWTH Aachen, which was founded 150 years ago. I received a bar of chocolate to commemorate the occasion. It was delightful.

Our discipline of operations research, which became a discipline after the second World War, did not exist 150 years ago. My department of Industrial and Systems Engineering celebrated its 50th anniversary in 2016.

I wonder when the first OR program was founded? Perhaps operations research was offered as a track in another degree such as math or engineering instead of being a standalone degree, as it generally is now. When was the first OR degree awarded and to whom? What about the woman to receive an OR degree? What about the first minority and minority woman to receive an OR degree? What were other important “firsts” for the discipline? If you know the answers, please post them in a comment or send me an email or a tweet.

If you are interested in learning more about the history of operations research, I recommend checking out the INFORMS YouTube channel on this subject.

Updates:

 

 

 


Achtung baby, my sabbatical is finally here!

This is my first sabbatical. I moved to UW-Madison when I was due for my first sabbatical, so my sabbatical clock started over again in 2013. Having a first sabbatical as a full professor is not the norm, but I’m taking advantage of the opportunity to spend a few months in Germany. I was delighted to be invited to visit RWTH Aachen’s fabulous OR Institute by Prof. Dr. Marco Lübbecke. I had the pleasure of visiting Aachen in 2014 when I gave a talk at the German OR Society Conference, and it is nice to return for a longer visit. My sabbatical is supported by a Fulbright Award.

I have some connections to Germany. I am of the granddaughter of German and Scottish immigrants. I also have some German heritage on the other side of the family. I took German in high school growing up, and I completed a minor in German in college. I studied at the Technical University of Darmstadt as an undergraduate student in pursuit of the minor. A picture of me at my college graduation 20 years ago at the University of Illinois is below. The university provided me with a stole with the colors of the German flag to wear at the graduation ceremony, which would be strange for Germans to do due to their uneasy relationship with patriotism and their flag.

My three children are with me on my sabbatical (ages 8, 12, and 15). They will take part in the German experience and (hopefully) learn German. So far, they like the many opportunities for discount shopping in Germany, recycling our water bottles in the machine at Aldi, public transportation, the many nearby playgrounds, and the thermal baths in Aachen. In the next few months, we will discover much more to like.

I will be blogging during my sabbatical. I will tag all sabbatical related posts to make the series of blog posts easy to find, and I’ll use the “PunkRockORinGermany” hashtag on twitter. Stay tuned for more information about my sabbatical and about sabbatical planning/logistics.


Punk Rock OR: A decade in review

As we mark the beginning of a new decade, I reflected on the last decade from a professional perspective. In the last decade I

  • Was nominated and won a National Association of Counties award with community partner Hanover County EMS for putting some of my research into practice
  • Was awarded a Young Investigator award and an NSF CAREER award
  • Had my third child
  • Was tenured at VCU
  • Was tenured at Wisconsin and moved to Madison
  • Had some personal setbacks and bounced back
  • Gave six keynote, plenary, and semi-plenary talks throughout the world in Germany, the Netherlands, and Brussels
  • Started new research in cybersecurity and disaster restoration
  • Served two terms on the INFORMS Board as VP of Marketing, Communications, and Outreach and notably took on big roles in the INFORMS advocacy efforts
  • Was awarded the INFORMS Impact Prize for my research in aviation security
  • Served one term as the Wisconsin CoE Assistant Dean for Graduate Affairs
  • Graduated my first four PhD students
  • Am engaged as ever with teaching at Wisconsin
  • Met many new wonderful colleagues
  • Collaborated with fabulous colleagues

I’m looking forward to the coming decade!


When should a football team attempt a two point conversion instead of an extra point? A dynamic programming approach.

On Sunday November 10, 2019, the Carolina Panthers were down 14 against the Packers early in the 4th quarter. They scored a touchdown, putting them down by 8, and they went for a two point conversion.  The two point conversion did not succeed. This has been the subject of debate, with journalists both applauding and criticizing the decision.

I created a dynamic programming model to determine whether or not to go for a 2 point conversion. The dynamic programming model is based on Wayne Winston’s book Mathletics, which is a fantastic introduction to sports analytics. The state captures the team with possession, the score differential when they obtain possession, and the number of remaining possessions. When there is one remaining possession, it is the last possession. When there are three remaining possessions, the team with the possession has two scoring attempts. Each possession ends in a touchdown, a field goal, or no score. I assume half of all games end in a tie. The probabilities I used are based on average team statistics. I do not model other decisions, such as whether to go for it on fourth down, although these could further improve a team’s win probability.

The slides are below.

Bottom line: teams should go for two points when they score a touchdown and they are down 10, 8, 3, or 2 or up by 1, 2, 4, or 5 (including the points from scoring the touchdown) near the end of the game. These conclusions hold when there are at least two additional possessions in the game.

If you have the last possession: go for 2 when a touchdown on this last possession puts you down by 2.

If you just scored a touchdown but your opponent will have the last possession: go for 2 when a touchdown puts you down by 2 or up by 1, 4, or 5. You normally will want to go for 2 when a touchdown puts you up by 2 except in this situation, because missing the extra point means your opponent could win with a field goal.

Carolina went for two when down by 8 after scoring a touchdown. According to my math, Carolina made the right choice. However, the best strategy does not guarantee a win nor does it drastically improve the win probability.

We can examine the decision in more detail. When down by 8 with four possessions to go (which matches up with when Carolina went for a two point conversion), a team has one of two choices:

  1. They could kick an extra point, which would give them a 11.3% win probability if successful (with probability 0.96) or a 7.9% win probability if not successful. Together, this yields a 11.2% win probability.
  2. They could go for a two point conversion. If they succeed (with probability 0.48), they would have a 18.3% win probability. Otherwise, they would have a 7.9% probability of winning if not successful. Together, this yields a 12.9% win probability.

There are four things to keep in mind:

  • Carolina improved their probability of winning by 1.7% by going for two.
  • A good process does not guarantee a good outcome.
  • Carolina was not likely to win using either approach.
  • Carolina could have further improved their win probability by considering other decisions (who is playing, which plays are called, and whether to go for it on fourth down).

My conclusions are summarized in the chart below. For more reading: Benjamin Morris of 538 wrote an article about when to go for two here. My analysis is consistent with his, although we make different comparisons.

When to go for a two point conversion in NFL football

 


data science isn’t just data wrangling and data analysis: on understanding your data

I have often heard rules of thumb, such as data science/analytics is 80% data cleaning/wrangling and 20% data analysis. I’ve seen various versions of this rule of thumb that all have two elements in common:

1) They assign various percentages of time to data cleaning and analysis, with the time allocated to cleaning greatly outweighing the time allocated to analysis.

2) Time is always partitioned into these two categories: cleaning and analysis.

I want to introduce a third category: understanding the data. This is a critically important part of doing data science and analytics. I acknowledge that many data scientists understand their data quite well, so I am not not criticizing the entire data science community. Instead, I want to point out and discuss the rich tradition of understanding the data that we have fostered in operations research and highlight the importance of this often overlooked aspect of working with data and building data-driven models. As an applied optimization researcher, I believe that data collection, understanding the data, problem identification, and model formulation are critical aspects of research. These topics are important training topics for students in my lab. To solve a data-driven model, we need to understand the data and their limitations.

Here are a few ways I have made efforts to understand the data:

  • I have asked questions about the data to subject matter experts (who shared the data with me).
  • I have done ride-alongs and observed service providers in action to see how they collect and record data as well as how they interpret data categories.
  • I have read manuals that describe data sets.
  • Summary statistics and other analytical tools shed light on distributions and processes that produce the data.
  • Disseminating research findings often results in good questions about the data sources from audience goers, which has improved my understanding of the data.
  • I have read other papers related to my problem that describe how data are collected in other settings.

Understanding the data has helped me understand the data’s limitations and apply the data meaningfully in my research:

  • Data sets often have censored data. What is not included in the data set may be critically important. There is no way to know what is not in a data set unless I understand how it was collected.
  • Some data points are nonsensical or misrecorded (e.g., 8 hour ambulance service times). Others are outliers and are important to include in an analysis. Understanding how the data were recorded help to ensure that the data are used in a meaningful way in the application at hand.
  • Some data points are recorded automatically and others are recorded manually. Both kinds can be high quality or low quality, depending on the setting, employee training, and the tool for automatic collection.
  • Understanding the data is a first line of defense when it comes to data and algorithm bias. Most data sets are biased in that they are not fully representative of the target population or problem, and understanding these biases can help prevent building models that are discriminatory and/or not effective when it comes to the problem at hand.
  • Understanding what data are not included in a data set has resulted in me asking for additional sources of data for my research. Sometimes I have been able to get better data if I ask for it.

Without understanding the data, the analysis could be a matter of garbage in, garbage out

This post covers just one of many issues required for avoiding algorithm bias in applying data science/analytics. Colleagues and I shared some of these additional thoughts with Cait Gibbons, who wrote an excellent article about algorithm bias for the Badger Herald. Read her article for more.