Math Internal Assessment

Comparing the rate at which a popsicle and a scoop of ice cream

melt

Syllabus: Maths Analysis and Approaches Higher Level

Table of Contents
Introduction

Aim

Optimizing Different Shapes of the Ice Cream

Sphere

Cylinder

Experimental Procedure

Data Collection

Modelling and Mathematical Manipulation of Result

Linearizing Logistic Graphs

Finding the R value

Residual Sum of Squares

Logistic Model of Popsicle Ice Cream

Logistic Model of Scoop Ice Cream

Determining the Later Melting Rate

Analysis & Conclusion

Work Cited

Appendix
Introduction
The inspiration of this mathematical investigation comes from an essential aid that helps me get through stressful

times and one of my favorite snacks - ice cream. Whenever my friends and I get stressed from the workload in

school we would always go out for some ice cream to take a break and clear our minds. It would always be

frustrating when I would have my ice cream while working only to find out that the ice cream has melted after a

period of time. This would also irritate me as I could not enjoy the entire ice cream as it would always end up

melting. Therefore, when choosing my math IA topic, I realized that I am able to apply my math skills to investigate

something that is highly relevant to my life - ice cream. I also thought how interesting it would be to connect

calculus math and modeling something that I love.

After coming up with this IA idea, I was very excited to conduct an experiment of finding out how fast an ice cream

would melt. Doing some research prior to the experiment I found out that different shapes of ice cream had different

melting times therefore I have decided to compare the rate of melting of a popsicle ice cream and a scoop of ice

cream to determine the optimal ice cream for me to have whilst doing work.

With both ice creams being the same weight of 100 grams. I predict that the popsicle ice cream would melt slower

than the scoop ice cream due to the fact that popsicle ice cream contains water whereas a scoop of ice cream

contains milk. Milk is denser than water therefore I predict that it would take a longer time to melt. After conducting

this investigation, I would be able to figure out which ice cream takes longer to melt. I also predict that both the ice

cream would melt at a linear rate.

Aim
The aim of this investigation is to compare the melting rates between a popsicle and a scoop of ice cream. I would

also be looking at the shapes of the ice cream and how they are able to factor into the melting time of the ice cream.
I intend to use a scatter plot on a graphing software, Desmos, to model the relationship of the ice cream as well as

calculus - differentiation and integration. As I would be modeling the melting time of the ice cream, I will be

collecting primary data.

Optimizing Different Shapes of the Ice Cream

Sphere
"
The surface area of the sphere can be represented as 4𝜋𝑟 ! and the volume can be represented as # 𝜋𝑟 # . In order to

differentiate this, we would have to use implicit differentiation due to the fact that the variable r is on both sides of

the equation. As per my hypothesis, the rate of change in volume is directly proportional to the surface area. This is

because, if the surface area is greater, there will be greater exposure to surrounding. More heat will be on the surface

which will result in a decrease melting time.

𝑑𝑣
µ 𝑆𝑢𝑟𝑓𝑎𝑐𝑒 𝐴𝑟𝑒𝑎
𝑑𝑡
"
Since the volume of sphere is # 𝜋𝑟 # and its surface area is 4𝜋𝑟 !
4
𝑑(3 𝜋𝑟 # )
= 4𝜋𝑟 ! ( −𝑘 )
𝑑𝑡
Differentiating the left side with respect to 𝑡
𝑑𝑟
4𝜋𝑟 ! .
= 4𝜋𝑟 ! ( −𝑘 )
𝑑𝑡
𝑑𝑟
= −𝑘
𝑑𝑡
Using separable differential equation and integrating both sides
𝑑𝑟 = −𝑘(𝑑𝑡)

6 𝑑𝑟 = −𝑘 6 𝑑𝑡

𝑟 = −kt + c
𝑐 in the equation represents the initial radius of the sphere.

This equation can show that the radius will decrease over time, and since the radius will decrease, the surface area

will also decrease. This should make the rate of melting slower over time as the exposed area is lesser.

Cylinder
The surface area of a cylinder can be represented by 2𝜋𝑟ℎ + 2𝜋𝑟 ! and the volume can be represented as 𝜋𝑟 ! ℎ. In

order to differentiate this, we would also use implicit differentiation due to the fact that the variable r is on both

sides of the equation.

𝑑𝑣
= −𝑘(𝑆𝑢𝑟𝑓𝑎𝑐𝑒 𝐴𝑟𝑒𝑎)
𝑑𝑡
𝑑𝑣
(𝜋𝑟 ! ℎ) = −𝑘(2𝜋𝑟 ! + 2𝜋rℎ)
𝑑𝑡
$% '% '( '% ')
Using partial derivatives: $& = >'( × '& @ + > ') × '& @
𝜕𝑣 𝜕𝑣
= 2𝜋rh and = 𝜋𝑟 !
𝜕𝑟 𝜕ℎ
𝜕𝑟 𝜕ℎ
F2𝜋rh × G + F𝜋𝑟 ! × G
𝜕𝑡 𝜕𝑡
𝜕𝑟 𝜕ℎ
𝜋r HF2h × G + F𝑟 × GI = (2𝜋𝑟 ! + 2𝜋rℎ)(−𝑘)
𝜕𝑡 𝜕𝑡
𝜕𝑟 𝜕ℎ
F2ℎ × G + F𝑟 × G = (2r + 2ℎ)(−𝑘)
𝜕𝑡 𝜕𝑡
𝜕𝑟 𝜕ℎ
F2ℎ × G + F𝑟 × G = −2𝑘r − 2𝑘ℎ
𝜕𝑡 𝜕𝑡

Therefore, comparing the terms with it’s like terms:

𝜕𝑟
!2ℎ × ' = −2𝑘ℎ
𝜕𝑡
𝜕𝑟
! ' = −𝑘
𝜕𝑡
𝜕𝑟 = (−𝑘)(𝜕𝑡)
𝑟 = −𝑘𝑡 + 𝑐
𝑐 in the equation represents the initial radius of the cylinder.
𝜕ℎ
!𝑟 × ' = −2𝑘𝑟
𝜕𝑡
𝜕ℎ
! ' = −2𝑘
𝜕𝑡
𝜕ℎ = (−2𝑘)(𝜕𝑡)
ℎ = −2𝑘𝑡 + 𝑐
𝑐 in the equation represents the initial height of the cylinder.

Experimental Procedure
To collect primary data for this investigation, I would suspend the popsicle ice-cream and scoop ice cream over a

bowl on top of a weighing scale and allow it to melt at room temperature as shown below.

Data Collection
Once I have collected the data, I will create a scatter plot on Desmos to model the curve that would best fit the

scatter plot. After obtaining the curve, through differentiation I would determine the rate of change and would then

determine when the change is the greatest to understand when each ice-cream should be finished.

Table #1
Data for popsicle Ice cream Data for sphere Ice cream
Time (Minutes) Weight of melted ice Time (Minutes) Weight of melted ice
cream (grams) cream (grams)
2:00 2.1 2:00 0.5
4:00 4.2 4:00 1.3
6:00 8.3 6:00 2.2
8:00 14.2 8:00 3.4
10:00 21.3 10:00 5.5
12:00 31.4 12:00 10.3
14:00 53.4 14:00 14.3
16:00 70.2 16:00 40.2
18:00 80.5 18:00 79.3
20:00 90.5 20:00 91.3
 22:00 95.6 22:00 96.4
24:00 97.2 24:00 98.3
26:00 98.5 26:00 99.1
28:00 98.9 N/A N/A
30:00 99.3 N/A N/A

Modelling and Mathematical Manipulation of Result

Linearizing Logistic graphs
General Form of Logistic Graph
𝐿
𝑓(𝑥) =
1+ 𝑒 *+(-*.)
Symbol/Variable Quantity Represented
𝑓(𝑥) Output of the function
L Curves maximum value
k Upper limit of the summation
x Real number (Constant)
c X-Value of the sigmoid midpoint

To determine the equation of the logistic graph I will determine the values for the variables L, k and c. The original

weight of the ice cream before it started melting is 100 grams. As L in the logistic curve represents the maximum

limit that the graph can reach the L value would be 100. The c value is the midpoint of all the x values therefore in

this case it would be 16. Finally, to determine the variable k I would substitute any point on the graph from table #1

for the popsicle ice cream as can be seen below.

𝐿
𝑓(𝑥) =
1 + 𝑒 *+(-*.)
Substitute point from Table #1 for popsicle ice cream. Values (x,y) (12, 31.4)
100
31.4 =
1 + 𝑒 *+(0!*01)
(31.4) × (1 + 𝑒 "+ ) = 100
(1 + 𝑒 "+ ) = 3.185
(𝑒 "+ ) = 2.185
Apply natural logarithm on both sides
4𝑘 = ln (2.185)
𝑘 = 0.195
By substituting all the values L, k and c the logistic function equation is determined.
100
𝑓(𝑥) =
1+ 𝑒 *2.045(-*01)

Substituting points to get the equation created a graph that was way off the scatter plot. I thought substituting

random points would help me achieve the equation as it is what I was taught to do for quadratics, however it is clear

that the method does not work for logistic graphs. Hence, I decided to use another method. In this method, I will use

the RSS method to find the best fit curve. Since this is a logistic function, I will first be linearizing it. Then with the

help of partial derivatives I will find the minimum error between the y values.

𝐿
𝑓(𝑥) =
1+ 𝑒 *+(-*.)
𝐿
𝑦 =
1 + 𝑒 *+(-*.)

𝑦S1 + 𝑒 *+(-*.) T = 𝐿
𝐿
= 1 + 𝑒 *+(-*.)
𝑦
𝐿
= 1 + 𝑒 *+- × 𝑒 *+.
𝑦
Taking 𝑒 *+. 𝑎𝑠 𝐵 as it is a constant, I get:
𝐿
− 1 = 𝐵𝑒 *+-
𝑦
𝐿−𝑦
= 𝐵𝑒 *+-
𝑦
I will be using natural log to simplify the exponent:
𝐿−𝑦
ln F G = ln (𝐵𝑒 *+- )
𝑦
𝐿−𝑦
ln F G = ln (𝑒 *+- ) + ln (𝐵)
𝑦
𝐿−𝑦
ln F G = −𝑘𝑥 + ln (𝐵)
𝑦
6*7
This equation is now in the form of a linear equation where y-axis can be represented by ln > 7
@ and the y-intercept

will be ln (𝐵).

Therefore, the linearized table looked like:

Table #2
 Time Weight Natural Logarithm
( Minutes ) (Grams) of the weight of the
ice cream
𝒙𝒊 𝒚𝒊 100 − 𝒚𝒊
ln F G
𝒚𝒊
2 2.1 3.84
4 4.2 3.13
6 8.3 2.4
8 14.2 1.8
10 21.3 1.31
12 31.4 0.78
14 53.4 -0.14
16 70.2 -0.86
18 80.5 -1.42
20 90.5 -2.25
22 95.6 -3.08
24 97.2 -3.55
26 98.5 -4.18
28 98.9 -4.5
30 99.3 -4.95

Diagram #1 Linearized Scatter Plot

Once the scatter plot is obtained, I will be finding the r value to check the correlation such that the line of best fit is

indeed a good fit to the scatter plot.

Finding the R Value
The persons correlation coefficient (r) is a measure of how far data points are from the line of best fit. The formula
to achieve the r value is shown below.
∑( 𝑥9 − 𝑥̅ )( 𝑦9 − 𝑦[ )
𝑟=
\∑( 𝑥9 − 𝑥̅ )! ∑( 𝑦9 − 𝑦[ )!

Symbol/Variable Quantity Represented

𝑟 Correlation Coefficient
𝑥9 Values of the 𝑥 variable in the sample
𝑥̅ Mean of the values of the 𝑥 variables
𝑦9 Values of the 𝑦 variable in the sample
𝑦[ Mean of the values of the 𝑦 variables

−364.3218144
𝑟=
√1120 x 118.9898393

𝑟 ≈ −0.998

Diagram #2 Linearized Scatter Plot with Line of Best Fit

Residual Sum of Squares
The residual sum of squares (RSS) measures the residuals of a regression model but in this case the level of variance

in the error term. Using RSS, the amount of error remaining between the regression and the data set will be

determined.

Since the best fit line showed a strong correlation therefore, now I will find the equation for the best fit line. To do

so, I will compare the y-values of the scatter plot and the best fit line, I would then solve for its gradient and y-

intercept which will minimize the error (gap between the 2 y-values.)

Formula for Residual Sum of Squares :

:

𝑅𝑆𝑆 = b( 𝑦9 − 𝑓(𝑥9 ))!

9;0

Symbol/Variable Quantity Represented

𝑦9 The 𝑖&) value of the variable to be predicted
𝑓(𝑥9 ) Predicted value of 𝑦9
𝑛 Upper limit of the summation

The formula has been squared such that I can consider all the errors as a positive value. If I consider both, positive

and negative error then the sum of that can become zero which will not help me to solve this.
05

b( 𝑦9 − 𝑦)!
9;0

Replace y for (𝑚𝑥9 + 𝑏) as it is the equation for best fit line. the below calculation will help me to calculate the total
error between the y-values.
05

𝑒𝑟𝑟𝑜𝑟 = b( 𝑦9 − (𝑚𝑥9 + 𝑐))!

9;0

Hence to minimize the error, I will use partial derivative to differentiate the equation with respect to 𝑚 and 𝑏 as they
are the variables changing. The x-value and the y-values are taken as constants as they are just the values of
coordinates.

05
𝜕
(b( 𝑦9 − (𝑚𝑥9 + 𝑐))! )
𝜕𝑚
9;0
05
𝜕
b ( 𝑦 − (𝑚𝑥9 + 𝑐))!
𝜕𝑚 9
9;0
Using chain rule to simplify the equation
05

b(2( 𝑦9 − 𝑚𝑥9 + 𝑐)(−𝑥9 ))

9;0

Distributive law of algebra and equating it to zero as I am finding the minimum

05

b( − 𝑥9 𝑦9 + 𝑚(𝑥 ! 9 ) + 𝑐(𝑥9 ) = 0
9;0
05 05 05
!
− b 𝑥9 𝑦9 + 𝑚 b(𝑥9 ) + 𝑐 b(𝑥9 ) = 0
9;0 9;0 9;0

Hence I get the first equation for system of equation:

05 05 05

𝑚 b(𝑥9 )! + 𝑐 b(𝑥9 ) = b 𝑥9 𝑦9
9;0 9;0 9;0

As I am solving for two unknown, I will not find the second system of equation by partially differentiating 𝑏:
05
𝜕
b ( 𝑦 − (𝑚𝑥9 + 𝑐))!
𝜕𝑐 9
9;0
05

b 2(𝑦9 − (𝑚𝑥9 + 𝑐)) = 0

9;0

Dividing both sides by two to simplify the equation:

05 05 05

b 𝑦9 − 𝑚 b 𝑥9 − b 𝑐 = 0
9;0 9;0 9;0

Getting the second equation:

05 05 05

𝑚 b 𝑥9 + b 𝑐 = b 𝑦9
9;0 9;0 9;0

Therefore, to use and solve the above two equations to find 𝑚 and 𝑏, I will now find ∑05 ! 05
9;0(𝑥9 ) , ∑9;0(𝑥9 ) ,

∑05 05
9;0 𝑥9 𝑦9 𝑎𝑛𝑑 ∑9;0 𝑦9 as shown in table #3 below

Table #3
Time Weight Natural Logarithm Time Squared Time multiplied by
( Minutes ) (Grams) of the weight of the the natural
ice cream logarithm of the
𝒙𝒊 𝒚𝒊 𝒍𝒏(𝒚𝒊 ) (𝒙𝒊 )𝟐 weight of the ice
cream

(𝒙𝒊 ) × 𝒍𝒏(𝒚𝒊 )
2 2.1 3.84 4 7.68
4 4.2 3.13 16 12.51
6 8.3 2.4 36 14.41
 8 14.2 1.8 64 14.39
10 21.3 1.31 100 13.07
12 31.4 0.78 144 9.38
14 53.4 -0.14 196 -1.91
16 70.2 -0.86 256 -13.71
18 80.5 -1.42 324 -25.52
20 90.5 -2.25 400 -45.08
22 95.6 -3.08 484 -67.73
24 97.2 -3.55 576 -85.13
26 98.5 -4.18 676 -108.8
28 98.9 -4.5 784 -125.97
30 99.3 -4.95 900 -148.64
Sum of 𝒙𝒊 Sum of 𝒚𝒊 Sum of 𝒍𝒏(𝒚𝒊 ) Sum of (𝒙𝒊 )𝟐 Sum of
(𝒙𝒊 ) × 𝒍𝒏(𝒚𝒊 )
240 865.6 -11.67 4960 -558.73

The natural logarithm is used to multiply the y value in order to linearize the graph. (𝒙𝒊 )𝟐 in the equation is used to

solve for the residual best fit line with minimum error. By organising the values in a table, I would be able to

implement the RSS formula as well as solve for the sum.

Solving using simultaneous equations:

4960𝑚 + 240𝑐 = −558.73
240𝑚 + 15𝑐 = −11.67

Using substitution method where the second equation gets equated in terms of 𝑐:
−11.67 − 240𝑚
𝑐=
15
Substitute the value of c
−11.67 − 240𝑚
4960𝑚 + 240 F G = −558.73
15
4960𝑚 + 16(−11.67 − 240𝑚) = −558.73
4960𝑚 − 186.72 − 3840𝑚 = −558.73
1120𝑚 = −372.01
𝑚 = −0.332152
Substitute the value of 𝑚 to determine the value of 𝑐:
240(−0.332152) + 15𝑐 = −11.67
 15𝑐 = 68.047
𝑐 = 4.536
Comparing this linear equation to the logistic equation which was manipulated to resemble a linear function:
𝐿−𝑦
ln F G = −𝑘𝑥 + ln (𝐵)
𝑦

−𝑘 = 𝑚
−𝑘 = −0.332152
𝑘 = 0.332152

ln(𝐵) = 𝑐
ln(𝐵) = 4.536
𝐵 = 𝑒 ".5#1

Hence the Logistic function obtained is:

100
𝑓(𝑥) =
1+ 𝑒 *2.##!05!-=".5#1
Logistic Model of Popsicle Ice Cream

By conducting the same steps above for the scoop ice cream, I have come up with the logistic equation of the scoop
022
ice cream to be 𝑓(𝑥) = . The graph below shows the logistic function of the scoop of ice cream.
0=> !".$%$&'(.()
Logistic Model of Scoop Ice Cream

Determining the Later Melting Rate

To find out which ice cream has a later melting rate I would be finding the point of inflection for the logistic graph.
Below the logistic function for the popsicle ice cream will be used to demonstrate.
100
𝑓(𝑥) =
1+ 𝑒 *2.##!-=".5"
𝑓(𝑥) = 100(1 + 𝑒 *2.##!-=".5" )*0
Using power rule and chain rule to differentiate the equation
𝑑
= −(1)100(1 + 𝑒 *2.##!-=".5" )*! × (1 + 𝑒 *2.##!-=".5" )*!
𝑑𝑥
= −100(1 + 𝑒 *2.##!-=".5" )*! × (𝑒 *2.##!-=".5" ) × (−0.332)
𝑓′(𝑥) = 33.2(1 + 𝑒 *2.##!-=".5" )*! × (𝑒 *2.##!-=".5" )
To find the second derivative I will be using power rule and product rule to differentiate the equation. The product
rule will be shown below.
𝑑(𝑢𝑣) 𝑑𝑣 𝑑𝑢
=𝑢 +𝑣
𝑑𝑥 𝑑𝑥 𝑑𝑥
𝑓′(𝑥) = 33.2(1 + 𝑒 *2.##!-=".5" )*! × (𝑒 *2.##!-=".5" )
𝑢 = 33.2(1 + 𝑒 *2.##!-=".5" )*!
𝑑𝑢
= −66.4(1 + 𝑒 *2.##!-=".5" )*# × (𝑒 *2.##!-=".5" )
𝑑𝑥
𝑣 = (𝑒 *2.##!-=".5" )
 𝑑𝑣
= −0.332(𝑒 *2.##!-=".5" )
𝑑𝑥
𝑓"(𝑥) = (33.2(1 + 𝑒 !".$$%&'(.)( )!% × −0.332(𝑒 !".$$%&'(.)( )) − 66.4(1 + 𝑒 !".$$%&'(.)()!$ × (𝑒 !".$$%&'(.)( ) × (𝑒 !".$$%&'(.)( )
−0.332𝑥+4.54 −3
𝑓"(𝑥) = −11.02𝑒(.)( (𝑒0.664 − 𝑒 )2𝑒−0.332𝑥+4.543

To find the point of inflection I will be equating the second derivative to 0 to solve for 𝑥 .
−0.332𝑥+4.54 −3
𝑓"(𝑥) = −11.02𝑒(.)( (𝑒0.664 − 𝑒 )2𝑒−0.332𝑥+4.543
−0.332𝑥+4.54 −3
−11.02𝑒(.)( (𝑒0.664 − 𝑒 )2𝑒−0.332𝑥+4.543 =0
2.11" *2.##!-=".5" *2.##!-=".5" )*#
(𝑒 −𝑒 )(𝑒 =0
𝑥 ≈ 13.67
This means the latest melting point for the popsicle ice cream is at 13.67 minutes.
The same steps were repeated for the scoop ice cream and it was found that the latest melting point for the scoop ice
cream is at 9.52 minutes.

Analysis & Conclusion

During the investigation, it is seen that the scoop ice cream melts at a much faster rate compared to the popsicle ice

cream. Initially it takes a longer time for the scoop of ice cream to start melting but it finishes melting faster than the

popsicle ice cream as can be seen in table #1.

The popsicle and scoop ice cream are both logistic graphs. The logistic graph demonstrates that there is a fixed

number of resources used and it is unable to go past that limit. As both the ice cream weighs 100 grams, the limit of

the graph would be at 100. The outcome of the experiment was different from my hypothesis as I predicted that the

ice cream would melt at a linear rate which was not the case. The ice cream would melt slowly at first, as time

passed it would start to melt faster and faster until the entire ice cream was done melting. The models created have

high level of accuracy as the experiment done for both ice cream was conducted on the same day and the same time.

This means that the temperature of the room which the ice cream was melting in is kept at a constant. If the

temperature were to differ, it would have significant impact on the results because temperature is a major factor in

melting rates.

After conducting the investigation, I have come to a conclusion that the popsicle ice cream is a much suitable ice

cream to eat when doing my work. This is due to the fact that compared to the scoop ice cream it takes longer to
finish melting and has a later melting rate. It is clear when doing work, the optimal ice cream to choose is the

popsicle ice cream, in the future when I feel like eating some ice cream while doing school work, I will be sure to

get a popsicle ice cream instead of a scoop.

 Works Cited

Murry, Cierra. "Residual Sum of Squares (RSS): What It Is, How to Calculate It." Investopedia, 28 May

2022, www.investopedia.com/terms/r/residual-sum-of-

squares.asp#:~:text=The%20residual%20sum%20of%20squares%20(RSS)%20measures%20the

%20level%20of,your%20model%20fits%20your%20data. Accessed 3 Jan. 2023.

StudySmarter. "Residual Sum of Squares." StudySmarter,

www.studysmarter.co.uk/explanations/math/statistics/residual-sum-of-squares/. Accessed 3 Jan.

2023.

Turney, Shaun. "Pearson Correlation Coefficient (r) | Guide & Examples." Scribbr, 13 May 2022,

www.scribbr.com/statistics/pearson-correlation-coefficient/. Accessed 4 Jan. 2023.

"What is Ressidual Sum of Squares(RSS) in Regression (Machine Learning)." Youtube,

uploaded by Kindson The Tech Pro, 3 Apr. 2019, www.youtube.com/watch?v=3W8M0Qn7hi4.
Accessed 3 Jan. 2023.
Appendix