28 ISE Magazine | www.iise.org/ISEmagazine
Latin America is a region of many challenges but
also of innovation and opportunities. It is a region of
rapid urbanization, with 80% of its population (505
million people) living in cities within a diverse urban
It also has great concerns and challenges in terms
of social inequality gaps; Latin America is arguably the most
unequal region in the world. A recent study conducted from
366 cities in the region showed that 172 million (or 58% of the
population studied) live in areas that have air pollution levels
above the guidelines of the World Health Organization.
Another characteristic of the region is heavy trafc in its cit-
ies. In my city of Bogotá, Colombia, the increments in travel
time due to this factor alone is about 50%; a 60-minute trip
takes on average 90 minutes to complete. It is clear that citizens
in Latin America have a complex landscape. Although we have
learned to embrace chaos, the positive aspects of the region
offset these challenges. But for industrial engineers, this is an
invitation for improvement.
Sustainable development goals (SDGs) are a call for action by
all countries to promote prosperity while protecting our planet
and environment. The SDG 11 (The Global Goals for Sus-
tainable Development, globalgoals.org/11-sustainable-cities-and-
communities) focuses on making cities inclusive, safe, resilient
and sustainable. In this context, Latin America has a lot to do
to meet this goal by 2030, but also most of the countries in the
global south.
To illustrate these challenges, only half the worlds popu-
lation has proper access to public transit. More than half the
worlds population does not have proper access to public spaces
like parks. Air pollution causes a significant amount of untime-
ly deaths. In addition, during the pandemic, more than 90% of
COVID-19 cases have appeared in cities, putting even more
social and economic pressure on urban settings. A lot of ISE
wisdom could be put into service to make our world better.
On the bright side, several policies and programs have
emerged from Latin America and perhaps may spread to
other regions. We have systems that operate at a different
scale or under other realities that have pushed us to develop
creative solutions. In this article, I will cover challenges re-
lated to transportation, recreation in public spaces and air
quality that have been partially tackled using ISE techniques
in transdisciplinary teams. I will cover some of these stories
where the ISE viewpoint has enriched the solutions in non-
Bogotá, Colombia,
is a capital city that
serves as an example
of the challenges
and solutions toward
creating sustainable
urban environments
in Latin America.
Photo courtesy of O.L. Sarmiento
Lessons from Latin America
for sustainable, healthier cities
ISE teams create solutions for recreation, traffic, pollution problems
By Andrés Medaglia
Bogotá, Colombia,
is a capital city that
serves as an example
of the challenges
and solutions toward
creating sustainable
urban environments
in Latin America.
November 2021 | ISE Magazine 29
traditional settings for our profession.
Two key tools of ISEs are operations research and analytics,
which focus on solving complex problems. Both share the
idea that, through mathematical and computational models,
we support better decision-making by transforming data into
actionable advice. Descriptive models provide insight from his-
torical data, predictive models try to view what will happen
in the future and prescriptive models provide actions for deci-
sions. More importantly, ISEs with their systemic view and ho-
listic perspective derive methodologies by coupling, extending
and integrating different models to provide complete analytics-
based solutions to complex problems and systems.
To do so, it is important to articulate three pillars. First, it is
important to identify the problem, its context, the stakeholders
and the systems boundaries and its dynamics. Second, once
we understand the problem, we can choose and build the right
model. If there is no model that is a good fit for our problem,
sometimes we need to extend these models through research.
Third, as these models feed from data, it is necessary to extract,
transform and load data in a systematic way.
These analytics-based solutions often require us to devise
an ensemble of models and to explore multiple scenarios sys-
tematically, so programming is the glue that puts everything
together. Finally, once we solve our model, we need to go back
and interact with the stakeholders through a visual and action-
able representation of the solution. So let us see how we have
put these ISE tools to work.
Expanding recreation opportunities
In this case study, we worked in a public space program called
the Recreovía program with the Institute of Sports and Recre-
ation of the city of Bogotá (IDRD). The Recreovía program
provides physical activity classes in parks, mainly for vulnerable
communities. This type of program is present in nine countries
in Latin America with more than 350 programs.
As the program grows, it is necessary to identify the best lo-
cations and schedules to expand the Recreovía hubs through-
out the city. This challenge involves assigning public resources
efficiently (A robust DEA-centric location-based decision
support system for expanding Recreovía hubs in the city of
Bogotá (Colombia),” Sepideh Abolghasem, Felipe Solano,
Claudia D. Bedoya, Lina P. Navas, Ana Paola Ríos, Edwin A.
Pinzón, Ands L. Medaglia and Olga L. Sarmiento, Interna-
tional Transactions in Operational Research, 2019).
To respond to this challenge, we developed RecreoBOG,
a decision support system for the IDRD. RecreoBOG uses
geographical data and information on locations, areas, theft
records, schedules and historic attendance. At the descriptive
layer, it is possible to visualize information through descrip-
tive models. In this case, we built a data envelopment analysis
model to classify programs according to their efficiency. At the
predictive layer, we built a model based on random forest to
estimate the number of potential participants in the new hubs.
At the prescriptive layer, to select the best sites to expand the
program, we used robust optimization models that maximize
A Recreovía hub in Bogotá provides physical activity classes in parks, mainly for vulnerable communities.
Photo courtesy of J.D. Pinzón
30 ISE Magazine | www.iise.org/ISEmagazine
Lessons from Latin America for sustainable, healthier cities
the number of participants in the program, taking into account
the limited budget, that at most one schedule is selected per
hub and that the sites are widespread throughout the city to
adequately cover its population.
It is worth emphasizing that these analytics models are inte-
grated – the output of one model is the input for another. ISEs
are able to build these integrated models to solve these complex
problems. In 2017, the IDRD used RecreoBOG for its expan-
sion plan. Our team created different scenarios and conducted
a “what if ” analysis with IDRD officials. Our team recom-
mended a set of 19 Recreovía hubs that consistently appeared
in the solutions of our prescriptive analysis.
Of course, things are not always smooth and crisp as the
output of our models. Considering new budget realities and
gathering input from the community and local policymakers,
the IDRD decided to open 10 new hubs out of the 19 initially
recommended. After the implementation of the new hubs, at-
tendance steadily increased. Total attendance data among all
hubs from February to May 2017 showed a compound month-
ly growth of 28% and an average per event attendance growth
of 7%.
Measuring cyclists’ trafc stress
In this second study, we developed a methodology to estimate
the level of traffic stress of cyclists in Bogotá. This work was
funded by the Wellcome Trust of the UK and is part of a larger
project called SALURBAL (a Spanish acronym for Urban
Health in Latin America), which has a main goal of making
cities healthier, more equitable and environmentally sustain-
Bogotá has the largest network of bikeways in Latin Ameri-
ca, with more than 550 km (more than 341 miles) of dedicated
bikeways. Cyclists use this bike network, combined with the
mixed-use road network, to complete their trips. In 2019, bike
trips accounted for 7% of the 13.3 million daily trips in the city.
It is important for the city to assess the level of stress that a cy-
clist experiences based on the physical and functional attributes
of the streets. At the end, our city would like to know how to
invest in infrastructure to have a bike network with lower stress
levels for cyclists.
But first, how do we measure the level of trafc stress (LTS)?
The LTS is an indicator that classifies the road segments ac-
A bike path in Bogotá. The Colombian capital has the largest network of bikeways in Latin America, with more than 550 km
(more than 341 miles) of dedicated bikeways. Cycling trips accounted for 7% of the 13.3 million daily trips in the city in 2019.
Photo courtesy of C. Fernández
It is clear that citizens in Latin
America have a complex landscape.
Although we have learned to embrace
chaos, the positive aspects of the
region offset these challenges. But
for industrial engineers, this is an
invitation for improvement.
November 2021 | ISE Magazine 31
cording to the stress experienced by
cyclists. To illustrate, if cars pass by
your side, you might feel threatened
and in danger. But if you ride on a
segregated bike path, you might feel
a bit more secure. These experiences
create different levels of stress.
Under the LTS umbrella, depend-
ing on the characteristics of the road
segment and its functional informa-
tion, the street is classified as one of
four levels. Typically, LTS 1 are streets
that are the least stressful roads, even
suitable for kids. On the other end of
the spectrum, LTS 4 are streets gener-
ally more stressful and are suitable just
for fearless cyclists.
Although LTS methodologies
have been developed and applied to
small- and medium-sized cities (such
as Portland, Oregon, in the U.S.), our
challenge was to extend the method-
ology to estimate LTS for a city of the
size of Bogotá. Our analytics-based
methodology starts with variable se-
lection. After discussing with trans-
portation experts, urban planners, and
public health experts in our group, we identified two groups of
variables: physical and functional. The physical variables were
road width, number of lanes, presence of cycling infrastructure
and heavy vehicles; the functional variables were vehicle speed,
traffic density, traffic flow and congestion.
Once we calculated the variables for all the road segments,
we used cluster analysis to identify the road segments that are
similar to each other in terms of the eight variables (“Level
of traffic stress-based classification: A clustering approach for
Bogotá, Colombia, Colombia,” Jorge A. Huertas, Alejandro
Palacio, Marcelo Botero, Germán A. Carvajal, Thomas van
Laake, Diana Higuera-Mendieta, Sergio A. Cabrales, Luis A.
Guzman, Olga L. Sarmiento and Andrés L. Medaglia, Trans-
portation Research Part D: Transport and Environment, 2020). Be-
cause Bogotá has about 170,000 road segments, the clustering
algorithm takes a long time to run. For this reason, we had to
adjust the classification algorithm to make it faster and scalable
to the size of the city. We chose a representative area of the city,
called a locality, and ran the clustering algorithm with the road
segments of that single area.
After classifying the road segments into one of the four clus-
ters (LTS low, LTS medium, LTS high and LTS extremely
high), we trained a multinomial logistic regression to predict
the probability of a new segment to belong to one of the four
stress levels. With these probabilities, we classified the rest of
the road segments in the city. Finally, based on this two-step
classification, we developed a dashboard (Figure 1) to visualize
these results. There we can see the LTS for a given neighbor-
hood and that the busier, wider streets tend to be LTS extreme-
ly high (red) and the residential streets within the neighbor-
hoods tend to be LTS low (green).
Our LTS methodology provide useful information for city
planning. We tested the hypothesis that more accidents occur
on segments classified as LTS high and LTS extremely high.
We overlapped the LTS layer with official reports of fatal and
nonfatal collisions. As expected, the fatal and nonfatal accidents
increased as the LTS level increased. Thus, we provided city
officials strong evidence to help them find resources to lower
the level of stress of the road network. Also, this methodology
allows city ofcials to efficiently predict the impact of planned
interventions on the LTS-based classification. When a new
segment is planned, now we can anticipate how much stress it
will impose on cyclists.
Measuring the multidimensional impact
of the bicycle transport system
A third case focuses on the multidimensional impact of cycling
as an active transportation mode that grew significantly as an
alternative in the COVID-19 pandemic. This project is spon-
sored by Engineering X, an international collaboration funded
Measuring cyclists’ road stress
A dashboard shows the level of traffic stress (LTS) classification in the Ciudad Salitre
neighborhood of Bogotá.
32 ISE Magazine | www.iise.org/ISEmagazine
Lessons from Latin America for sustainable, healthier cities
by the Royal Academy of Engineering and Lloyds Register
Foundation (“Engineering X. Safer Complex Systems,Royal
Academy of Engineering, case studies).
On March 20, 2020, the city of Bogotá entered a strict lock-
down due to the global COVID-19 pandemic. The Mobility
and Health Secretariats looked into innovative ways to guaran-
tee mobility throughout the city while reducing the capacity
of the public transport system complying with social distancing
recommendations. Based on the city’s established bike culture,
local government targeted temporary bike paths as the primary
solution, and by June 2020, it added 117 km (72.7 miles) of
temporary bike lanes on top of the 550 km of existing dedi-
cated bikeways. This expansion was planned and conceived by
mirroring the corridors of the bus rapid transit, adding cycling
infrastructure in areas with high volume of cyclists and adding
more connectivity to longer bike trips.
With this context in mind, this project focuses on Bogotás
bicycle transport system and the measures taken by the city to
manage the crisis. We evaluated the potential impact of the
temporary bikeways in terms of safety, health, efciency and
flexibility. We designed a systems analytics methodology that
combines systems modeling and analytics to analyze the bicycle
transport (complex) system.
The systems analytics methodology is an innovative meth-
odology for evaluating systems that combines actors’ knowl-
edge with data collected from possible interventions of the sys-
tem. It uses analytics (descriptive, predictive and prescriptive)
models to measure the system regarding specific performance
indicators. The results of this case study are available at the
Safer Complex Systems-Engineering X site, raeng.org.uk/safer-
Tackling Bogos air quality
The fourth and last case focuses on air quality at a larger scale
for Bogotá. This project, funded by the city, brought together
researchers from environmental and industrial engineering, the
business school and the school of economics to solve a critical
environmental and health problem for the city.
Rapid urbanization has a direct impact on air quality and
health-related problems. In 2015, the total cases of mortality
attributable to air pollution reached 3,000 deaths, about 10.5%
of the city’s total deaths. The estimated cost due to air pollution
More than 10% of Bogotá deaths were attributed to air pollution in 2015, with an estimated cost to the city of $1.5 billion-plus.
ISEs with their systemic view
and holistic perspective derive
methodologies by coupling, extending
and integrating different models to
provide complete analytics-based
solutions to complex problems and
November 2021 | ISE Magazine 33
has surpassed $1.5 billion (Ambient fine particulate matter in
Latin American cities: Levels, population exposure and associ-
ated urban factors,” Nelson Gouveia, Josiah L. Kephart, Iryna
Dronova, Leslie McClure, José Tapia-Granados, Ricardo Mo-
rales-Betancourt, Andrea Cortínez-ORyan, José Luis Tex-
calac-Sangrador, Kevin Martinez-Folgard, Daniel Rodriguez
and Ana V. Diez-Roux, Science of The Total Environment, 2021).
We were part of a multidisciplinary project that proposed
a 10-year decontamination plan for the city. The project pro-
vided a prescriptive optimization-based integrated assessment
modeling tool to support a comprehensive urban air quality
policy in Bogotá (“Optimization model for urban air quality
policy design: A case study in Latin America,” Jorge A. Sefair,
Mónica Espinos, Eduardo Behrentz and Andrés L. Medaglia,
Computers, Environment and Urban Systems. 2019). It provided
city officials with a portfolio of mitigation actions for air qual-
ity policymaking including implementation cost, pollution
goals, interdependencies between alternatives and managerial
and regulatory constraints. It finally supported the design of an
air quality plan to accomplish objectives on emission levels at
the city level in a 10-year horizon.
The output of our model was a portfolio of mitigation ac-
tions and the optimal timing for execution of these strategies,
given a set of constraints like budget, pollutant goals and regu-
latory constraints. This plan takes into account early and tardy
dates for implementing each mitigation strategy, the capacity
of the city to implement the policies, synergies between the
strategies, economic costs and environmental impact (the re-
duction of pollutants). A key feature of this model is the fact
that for each mitigation strategy, it is necessary to select a mode
of implementation. These different modes have different re-
source consumption and more importantly, they have different
environmental inputs.
A health benefit analysis of the proposed air quality plan es-
timated the health impacts and costs for the optimal plan (for
2010-2020), including 17,600 avoided deaths associated with
long-term exposure to PM
(solid and liquid particulate mat-
ter less than 10 micrometers in diameter); a decrease in 1,600
cases of child mortality; 30,100 avoided hospitalization cases
linked to respiratory diseases in children under age 5; and a
reduction in 11,900 hospitalization cases due to cardiovascular
Implementation of the plan has encountered multiple ob-
stacles. Some measures were sanctioned as decrees; others were
not applied. But all in all, it had partial impact on improving
the air quality in the city. Since 2010, the fraction of days per
year where PM
concentration exceeded the national standard
of 15 micrograms per cubic meter has decreased, showing a
significant improvement in air quality. Again, we cannot argue
that it is just the result of implementing the prescriptive model,
but it is certainly one of the multiple causes of this positive ef-
ISEs meeting the challenges
Latin America is a region full of challenges and opportuni-
ties. I have put an emphasis on the problems we need to solve,
but there are many more positive aspects. In the region there
are a lot of ISEs who are passionate about solving these prob-
lems. Throughout these projects, working in multidisciplinary
teams, we have seen how professionals from other disciplines
have seen the power of ISE and have embraced our tools to
make them their own, allowing us to shift to truly transdisci-
plinary research. These sustainability problems are a constant
call for innovation with other disciplines.
Solutions for complex problems often demand the integra-
tion of multiple models: descriptive, predictive and prescrip-
tive. As ISEs, we are called to be analytics solution architects
with a systemwide perspective. To impact our communities,
it is necessary to engage policy and community actors in the
co-creation of these solutions using systems-thinking method-
ologies. This takes time but builds a solid ground for long-term
Andrés Medaglia is a full professor of industrial engineering at Univer-
sidad de los Andes in Bogotá, Colombia, an IISE University Partner.
He has nearly 20 years of experience conceiving, designing and de-
veloping analytics models that support strategic and operational deci-
sions in a wide range of domains, including transportation and urban
systems. His research has led to over 60 peer-reviewed publications and
over 180 oral presentations in analytics and operational research. He
serves in the editorial boards of Transportation Science, Comput-
ers and Operations Research, the European Journal of Indus-
trial Engineering and TOP, the ofcial journal of the Spanish OR
and Statistics Society. He has received several awards, more recently the
Glover-Klingman Prize (2020), the INFORMS/TSL Presidents
Service Award (2017) and the IFORS Prize for OR in Development
(2017). He is an IISE member and has served as vice president of
Central/South America for IISE. Contact him at andres.medaglia@
Solutions featured in 2021
Annual keynote
Author Andrés Medaglia presented a summary of his team’s
findings in his keynote speech at the virtual 2021 IISE Annual
Conference & Expo. His presentation and other conference
content remains accessible to attendees with their site login at
iise.org/Annual2021. To register and learn more about the 2022
conference scheduled for May 21-24 in Seattle, Washington,
visit iise.org/Annual.