Teaching and Learning Mathematical Modelling

with Technology
Keng-Cheng Ang
[email protected]
National Institute of Education
Nanyang Technological University
1, Nanyang Walk, Singapore 637616

Abstract: In the last few decades, there have been abundant discussions among mathematicians and mathematics
educators on promoting mathematical modelling (a process of using mathematics to tackle real world problems) as a
classroom practice. Mathematics educators and curriculum planners have been advocating the teaching of
mathematical modelling in schools for some time now. Despite the consensus on its importance and relevance,
mathematical modelling remains a difficult activity for both teachers and learners to fully engage in. In this paper, we
examine some of these difficulties and discuss how technology can play a pivotal role in providing the essential support
to make mathematical modelling a more accessible mathematical activity amongst students. Through a series of
examples drawn from different fields and topics, we illustrate how a range of technological tools may be successfully
and efficiently utilized in modelling tasks. In addition, we discuss the need for an optimal use of technology to balance
between achieving the objectives of the tasks and attaining the goals of learning mathematics.

1. Introduction
Mathematical modelling may be loosely defined as a process of representing real world problems in
mathematical terms in an attempt to understand and find solutions to the problems. A mathematical
model can be considered as a simplification or abstraction of a (complex) real world problem or
situation into a mathematical form, thus converting the real world problem into a mathematical
problem. The mathematical problem can then be solved using familiar mathematical techniques.
The solution obtained is then interpreted and translated into real terms. Although there may be
several interpretations of mathematical modelling, the process of mathematical modelling may be
represented as a flow of events illustrated in Figure 1.1.

As depicted in the figure, we begin with a real world problem and we wish to find a real world
solution to this problem. This may be difficult to achieve directly in the real world. We thus make
an attempt to understand the problem, and then describe it in mathematical terms. At this stage, it is
often necessary to identify the variables in the problem and construct relationships between or
amongst these variables. Next, we develop a basic framework for the model. Here, assumptions
about the model may need to be made to keep the problem tractable and simple so that we are able
to solve the model using known methods.

Based on these assumptions, we construct a model, which could be a single equation, or a set of
equations, or a set of rules or simply an algorithm governing how values of the variables may be
found or assigned. This is the most crucial stage during which one would usually justify the
formulation of the model based on the real physical meanings of the variables in the problem. Very
often, model formulation is the most challenging stage for students (and teachers) as it requires
fairly high order thinking, inter-disciplinary knowledge and modelling experience.
 Real World Mathematical World

Real-world Mathematical Make

Problem Problem Assumptions

Model Formulation

Formulate
Equations

Real-world Interpret Solve

Solution Solutions Equations

Model
Interpretation
Compare Model Refinement
with data

Figure 1.1: The modelling process (adapted from Ang, 2006b)

Once a model is constructed, the next stage requires the modeller to find ways to solve the model,
using various mathematical techniques and tools. Very often, unless a model is particularly simple,
some kind of technological or computing tool will be necessary. One also often finds that there can
be a variety of ways of solving the same problem, making mathematical modelling a very enriching
mathematical experience. We then interpret the results or solutions of the model in the context of
the real world problem, and make attempts to compare the model solutions and any collected or
known data. Sometimes, we wish to refine the model by revisiting and revising our assumptions.

2. Technology and Mathematical Modelling

Despite its importance and relevance to the real world, mathematical modelling is generally not the
main approach to teaching and learning of mathematics in schools. For instance, although it has
been proposed that mathematical modelling can be introduced to in Singapore schools (Ang, 2001),
it was not until recently that such a suggestion was noticed by the local curriculum planners.

One reason could be the lack of readily available resources (lesson plans, modelling tasks, and so on)
for the teacher, notwithstanding a recent attempt to develop resources for local teachers (Ang, 2009).
Another is the teacher’s lack of experience in mathematical modelling, leading to a lack of
confidence and a general reluctance to embark on mathematical modelling in the classroom. At
times, there is concern that students may not be mathematically ready for the tasks that teachers
have painstakingly designed. As pointed out by Ang (2010), all these are stumbling blocks to an
otherwise enriching and exciting approach to learning and teaching mathematics.

Technology may be the bridge for the cognitive gap that hinders a student from carrying out a
modelling task. However, it should also be noted that technology should never replace the
mathematics, much less the teacher; it should be viewed as a timely, and sometimes temporary,
means of overcoming a difficulty.
The approaches to teaching mathematical modelling have been influenced by the development and
introduction of technologies such as graphing calculators and computer software (Ferrucci and
Carter, 2003). Many researchers and teachers have reported the successful use of technology in
introducing mathematical ideas through exploration and investigation. For instance, the use of a
spreadsheet to explore mathematical concepts has been discussed by Chua and Wu (2005) for a
secondary classroom, and by Beare (1996) at college level. Ang and Awyong (1999) reported that
the use of computer algebra systems such as Maple in some tertiary courses has been well received.
Not surprisingly, the use of technology continues to prevail in the mathematics classroom at all
levels.

In the next section, we discuss four examples. In each example, we illustrate how technology can
be successfully employed in the modelling task. In particular, the examples will show the role that
technology can play in bridging gaps in mathematical knowledge or skills, without diminishing the
student’s mathematics learning experience by any large extent.

3. Examples
3.1 Warming Up

One simple technological tool that can be used for capturing data for a modelling task is the data
logger and the accompanying software, LoggerPro 3 (Figure 3.1). For instance, a modelling task
could involve students using the temperature probe to capture the temperature of a cup of ice water
as it warms to room temperature. As shown in Figure 3.1(b), a typical data set shows the
temperature of water collected at 5-second intervals.

(a) Data logger (b) A screenshot of LoggerPro 3

Figure 3.1: Modelling the warming of ice water using a data logger and LoggerPro

The task could be to “develop” or “discover” a mathematical model to describe the process of
warming, in this case, of ice water. Of course, if we are aware of Newton’s law of cooling/warming,
we can immediately apply the law. However, the point of the task is to take the student through the
process of modelling.
Questions that one may ask students performing this task include:
(1) What are the factors (variables) that can influence of affect the temperature of the water?
(2) What happens near the beginning and near the end of the experiment?
(3) What assumptions do we need to make about the warming process?
(4) How quickly or slowly does the temperature change at different times?
(5) What can you say about the rate of change of the temperature? Write down a word equation
that describes the rate of change.
(6) Write down a differential equation that describes how the temperature changes with time.

Notice that these questions can be handled by students at different cognitive levels. For instance, a
pupil in the primary (elementary) school may be able to handle the more basic questions like (1), (2)
and (3), whereas a secondary (high school) or Junior College (pre-University) school student should
be able to tackle all the questions, including higher order ones like (4), (5) and (6).

From these questions, hopefully, students can be led to “discover” that the rate of warming (or
cooling) of an object is directly proportional to the difference between its temperature ( ) and that
of the surrounding ( ). Representing this as a differential equation, we obtain
(3.1)
where is some constant to be estimated from the data.

3.2 Queues

In our everyday life, we often experience queuing systems. Some examples include a bank-teller
service, drive-through service at a fast-food restaurant, printing jobs for a computer network, and so
on. How can we introduce a queuing system as a modelling activity in a mathematics classroom?

In a simple single-server queuing model, the three components involved are arrival process, service
process, and queue structure. A good example for students to think about is a queue at an automatic
teller machine or ATM. The ATM can be considered as a machine that serves one customer at a
time, on a “first in first out” basis. Customers arrive at the ATM randomly over time, wait their
turn in line, and spend a random amount of time at the machine before leaving.

ATM

Customers arriving at Customer

the server being served

Figure 3.2: A single-server queuing system

The modelling task is to construct a model that can simulate such a queuing system. Like before,
one can guide the student with questions like those in the previous example.

Here, two important variables are the arrival rate and service rate. Suppose we let be the number
of customers arriving per unit time, and represent the number of customers served per unit time.
Assuming that both customer arrivals and service are Poisson processes, then inter-arrival times and
service times may be generated using an exponential distribution.
To generate random numbers from an exponential distribution with parameter , we first generate a
random number from a uniform distribution in the interval , (which can be easily achieved in
a spreadsheet like Excel), and then compute
(3.2)

A simple simulation of this queuing model may be carried out on a spreadsheet such as MS Excel.
Figure 3.3 shows a screenshot of a typical simulation run. Pressing the “Calculate Now” button, F9,
generates another run.

Figure 3.3: Screenshot of an Excel worksheet used to simulate a simple queue

The above simulation is carried out as follows.

1) Columns B and D are random numbers chosen from an exponential distribution with
parameters and respectively using the method described earlier. These represent the inter-
arrival times (IAT) and service times (ST) respectively for each customer.
2) Column C records the actual arrival time (AT) by adding the contents of the previous cell to the
corresponding IAT (e.g. Cell C7 = C6 + B7).
3) The Finish Time (FT) in Column E is found by taking the larger of the sum of the previous FT
and the corresponding ST and the sum of the corresponding AT and ST (e.g. Cell E7 =
MAX(D7+C7, E6+D7)).
4) The total time spent by each customer (Column F) is the difference between the customer’s FT
and AT (e.g. Cell F7 = E7−C7).
5) The total wait time for each customer (Column G) is the difference between the customer’s
total time spent and ST (e.g. Cell G7 = F7 – D7).

Estimates for the parameters and may be obtained by experiment. One could video-record a
simple queue at an ATM, and estimate customer arrivals and service durations for each customer.
Average rates can then be calculated and fed into the simulation model. The simulation can also be
automated and improved with the use of macros or VBA programs written in Excel.
In constructing a model for the simple queue, we may consider questions such as:
(1) How likely is it that a customer will need to wait to be served?
(2) What factors affect this likelihood?
(3) What is the average wait time?
(4) What options are there to reduce this wait time?

Queuing Theory is not an easy concept for primary or secondary high school students to grasp.
However, queuing or waiting is a common experience in our everyday life. Using a simulation
model, we can demonstrate how mathematics can be used to study such a complex process.
Students will not need to understand the concept of Poisson process or exponential distribution until
they are ready. In the mean time, the cognitive gap can be bridged by technology, giving them the
opportunity to appreciate the power of mathematical modelling in an everyday life experience.

3.3 Falling rain

This example is inspired by a discussion on a modelling task, carried out by Oldknow (2003), who
illustrated how Geometer’s Sketchpad (GSP) may be used to study the motion of water spouting out
from a Singapore landmark called the Merlion. A photograph of the Merlion showing the trajectory
of water was used as the subject of study, and GSP, as well as Cabri Geometry (II Plus), was used
successfully to examine the geometry of the motion.

It often pours in Singapore, and this year, we have seen floods at various times of the year at
various places on the island. In fact, Orchard Road, a well known shopping district in the city state,
turned into “Orchard River” on one occasion.

At the National Institute of Education, we have had our fair share of flooding experience as well.
On one stormy day in June this year, rain water had gathered in the newly renovated terrace just
outside our staff offices, and eventually overflowed into the corridors and flooded the offices. The
estate and building management decided that the way to arrest this problem was to collect the water
coming down from the gutter and draining it off in the manner shown in Figure 3.4. We can import
the same image onto GSP and carry out a modelling task, as depicted in the screenshot in Figure 3.5.

Figure 3.4: Rain in the terrace

Based on the laws of particle motion, we can guess with reason that the trajectory of the water is a
parabola. In other words, if we place the origin O at the point where the water just leaves the gutter,
the function used to describe the trajectory is simply , where is a negative constant.
Fixing , and using the slider technique as suggested by Oldknow, we can vary the unit
distance of the coordinate system until the path of the water is as close to as we can.

One objective of this exercise could be to study the way rain shoots off the gutter, and evaluate the
effectiveness of the current method of solving the flooding problem. We can also use the image and
GSP to find the height of the roof where the gutter is, the angle at which the water hits the ground
and so on.

Figure 3.5: Modelling the path of rain water from the gutter using GSP

3.4 Disease Outbreak

In 2003, Singapore had to grapple with the outbreak of a deadly disease known as Severe Acute
Respiratory Syndrome, or SARS in short. In a short span of 70 days, 206 cases were reported and
of these, 31 lives were lost. Data for the SARS outbreak in Singapore are available in the public
domain and reproduced below (Table 3.1). These can serve as useful material for a modelling task
for students (Ang, 2003).

Given this set of data, students may be challenged to use a mathematical model for an epidemic to
explain or describe the process of the outbreak. One simple epidemic model that may be suitable is
the “S-I” model. In this model, the two compartments of populations are the susceptible individuals
(S) and the infected (I) individuals of a community. For simplicity, we may need to assume that the
community is closed.
 Table 3.1: Number of individuals infected with SARS during the 2003 outbreak
in Singapore (Heng and Lim, 2008)
Day Number Day Number Day Number Day Number Day Number
( t ) ( x ) ( t ) ( x ) ( t ) ( x ) ( t ) ( x ) ( t ) ( x )
0 1 15 25 29 101 43 163 57 202
1 2 16 26 30 103 44 168 58 203
2 2 17 26 31 105 45 170 59 204
3 2 18 32 32 105 46 175 60 204
4 3 19 44 33 110 47 179 61 204
5 3 20 59 34 111 48 184 62 205
6 3 21 69 35 116 49 187 63 205
7 3 22 74 36 118 50 188 64 205
8 5 23 82 37 124 51 193 65 205
9 6 24 84 38 130 52 193 66 205
10 7 25 89 39 138 53 193 67 205
11 10 26 90 40 150 54 195 68 205
12 13 27 92 41 153 55 197 69 205
13 19 28 97 42 157 56 199 70 206
14 23

Suppose we let be the number of infected (and hence infectious) individuals, and be the total
number of individuals in the community. Then, the S-I model reduces to the logistic equation

(3.2)

where is a positive constant representing the rate of transmission of the disease. Assuming that
the initial condition is given as , and using a standard method of solution (such as
separation of variables), the solution to Equation (3.2) may be written as

(3.3)

From Table 3.1, it is clear that , meaning that the outbreak had started with one
infected individual. In addition, it is assumed that the total number of individuals in this closed
community is . Although this assumption is open to debate, it is a necessary assumption if
we wish to use this model. With these values, what is needed to complete the model is a value for
the transmission rate, . Here is where the “Solver Tool” in MS Excel can be very useful. We first
define an “average error”,

(3.4)

where are the data values, are the values generated from the logistic model. In this case,
(the total number of data points available). Using the “Solver Tool” in MS Excel, we can
find a value for that minimises the average error, . In the present case, the minimum value of
is found to be 1.9145 when . Plotting the graph of the solution given in (3.3) with this
value of , it is clear that the model follows a similar trend as the data (see Figure 3.6), but it can be
improved. To improve the model, we observe that near the beginning and end of the outbreak, the
model appears to be fairly reasonable. However, between and , the model does not
seem to fit very well with the actual SARS cases.
 k= 0.1686 E= 1.9145

SARS Logistic Squared

Day
cases Model Error
0 1 1.0000 0.0000
1 2 1.1826 0.6682
2 2 1.3982 0.3621
3 2 1.6529 0.1205
4 3 1.9536 1.0950
5 3 2.3083 0.4785

70 206 205.6838 0.1000

(a) Minimizing error using a spreadsheet (b) Graph of SARS cases and solution from model
Figure 3.6. Using Excel’s Solver Tool in SARS outbreak model

Now, the logistic equation assumes a linear relationship between the fractional rate of change of
with . A more general model would be to assume that the fractional rate of change
of varies with for some real constant . We then apply the same procedure as
before to find values of and that will minimize the average error. The result is a modified or
generalized logistic model for the SARS outbreak and is shown in Figure 3.7.

(a) Modified logistic model with (b) Double logistic model with four parameters
and (See Ang, 2004 for details)
Figure 3.7: Refined models for SARS outbreak

It is clear from Figure 3.7(a) that the modified logistic model is an improvement over the logistic
model. Results can be further improved by using a “double logistic” model, as shown in Figure
3.7(b). Justifications for using a double logistic model and details on the approach can be found in
Ang (2004).
This example illustrates the concept of model refinement in mathematical modelling. Moreover, it
demonstrates the use of empirical data to estimate parameters, such as and . While has a
physical meaning, can be considered as a parameter in a deterministic model.

4. Conclusion
It has been suggested that mathematical modelling can provide a “unifying framework” for teaching
applied mathematics (Smith, 1996), without necessarily adding content to the curriculum. The way
in which using and applying mathematics is presented may need some redesigning. Mathematical
modelling could well be the approach to adopt as it provides opportunities to learn and apply
mathematics at the same time (Warwick, 2007). Given proper guidance and scaffolding, students
can indeed learn mathematical modelling and through the process, learn mathematics (Blum and
Ferri, 2009).

It is clear from the examples discussed that mathematical modelling can provide very rich learning
experiences in mathematics, and often contains inter-disciplinary elements. Problems can come
from some other discipline and the mathematics teacher can capitalize on these opportunities to
collaborate with other teachers in mathematical problem-solving. In addition, modelling exercises
set in a local context adds authenticity to the task and arouse greater interest amongst students. It is
for this very reason that the example on SARS model has received much attention and interest from
teachers and students in Singapore when it was first discussed.

It is equally clear, as illustrated in the examples, that the use of technology plays an important role
in mathematical modelling. Real life problems often involve real life data. These may need to be
collected and manipulated at times, and technology can make it possible and convenient even for a
young school pupil.

In some cases, technology can also help make the mathematics more accessible. For instance, in
the SARS example, the Solver Tool in Excel helps us find parameters for the model. Although it is
possible to work out the parameters by hand, the knowledge and skills required may be a little too
advanced for the intended learners. Rather than having to struggle with complicated or tedious
numerical computations, it may be better to use a tool so that one could focus on the model, the
application and the mathematics. The use of this technological tool thus bridges the gap, which the
student can fill in future.

As pointed out by Ang (2006a), when a mathematical modelling task is used as an approach to
learning mathematics, technology may be employed in to help the student to “do more with less
mathematics”. This includes exploring possible graphical solutions of the problem, performing
computational experiments or simulations in models, and manipulating or working with real data.

Using technology in the classroom for mathematical modelling activities can be a complex process.
It is not a simple matter of pushing or offering computers, calculators and software to school
children and teachers ad infinitum. One should be careful not to lose sight of the mathematics and
the spirit of mathematical modelling even when technology provides all the conveniences, and
sometimes, some of the answers.
References

Ang, K.C. & Awyong, P.W. (1999). The Use of Maple in First Year Undergraduate Mathematics.
The Mathematics Educator, 4(1), 87-96.
Ang, K.C. (2001). Teaching Mathematical Modelling in Singapore Schools, The Mathematics
Educator, 6(1), 62-74.
Ang, K.C. (2004). A simple model for a SARS epidemic, Teaching Mathematics and Its
Applications, 23(4), 181-188.
Ang, K.C. (2006a). Mathematical Modelling, Technology and H3 Mathematics, The Mathematics
Educator, 9(2), 33-47.
Ang, K.C. (2006b). Differential Equations: Models and Methods. McGraw-Hill, Singapore.
Ang, K.C. (2009). Mathematical Modelling in the Secondary and Junior College Classroom,
Singapore: Prentice Hall.
Ang, K.C. (2010). Mathematical Modelling in the Singapore Curriculum: Opportunities and
Challenges, In Proceedings of the Educational Interface between Mathematics and Industry Study
Conference, Lisbon, Portugal, pp. 53-61.
Beare, R. (1996). Mathematical Modelling Using a New Spreadsheet-based System, Teaching
Mathematics and Its Applications, 15(1), 120-128.
Blum, W. & Ferri, R.B. (2009). Mathematical Modelling: Can it be taught and learnt? Journal of
Mathematical Modelling and Application, 1(1), 45-58.
Chua, B. & Wu, Y. (2005). Designing Technology-Based Mathematics Lessons: A Pedagogical
Framework. Journal of Computers in Mathematics and Science Teaching, 24(4), 387-402.
Ferrucci, B.J. & Carter, J.A. (2003). Technology-active mathematical modelling, International
Journal of Mathematical Education in Science and Technology, 34(5), 663-670.
Heng, B.H. & Lim, S.W. (2003). Epidemiology and control of SARS in Singapore, Epidemiological
News Bulletin, 29, 42–47.
Oldknow, A. (2003). What a picture, what a photograph, Teaching Mathematics and Its
Applications, 22(3), pp. 141-144.
Smith, D.N. (1996). Mathematical Modelling. Teaching Mathematics and Its Applications, 15(1),
37-41.
Warwick, J. (2007). Some Reflections on the Teaching of Mathematical Modeling, The
Mathematics Educator, 17(1), 32-41.