Data Fitting — §1.3 & §3.

2 9

Data Fitting

Definition: A mathematical model is descriptively realistic if it is

deduced from a believable description of the system being modeled.

Example. Full moons. A full moon appears to occur every 29 days.

Let ML , MN be the dates of the last and next full moons. Is the
model
MN = ML + 29
descriptively realistic? Why?

Given a descriptively realistic model that gives a set relationship

that depends on a constant, how do we determine this constant?
Data Fitting — §1.3 & §3.2 10

Springs and Elongations

Example: Modeling Spring Elongation
m e
Take your favorite spring. Attach different masses.
50 1.000
How much does it stretch from rest? [Its elongation.]
100 1.875
When we plot the data, we get the following scatterplot. 150 2.750
200 3.250
Elongation !e"
Elongation of a Spring
250 4.375
10

8
300 4.875
6
350 5.675
4 400 6.500
2 450 7.250
0 100 200 300
Mass !x"
400 500 500 8.000
550 8.750
What do you notice?
Data Fitting — §1.3 & §3.2 11

Proportionality

When data seems to lie on a line through the origin, we expect the two
variables to be proportional; in this case, e = km for some constant k.
We need to find this constant of proportionality k.

So: Estimate the slope of the line. How?

1 Guesstimating
Elongation of a Spring
Elongation !e"
10

Mass !x"
0 100 200 300 400 500

2 Mathematically: Linear Regression / Least Squares

(For another day)
Data Fitting — §1.3 & §3.2 12

Fitting Gravity Data

Example. Modeling the dropping of a golf ball t x

Let’s use an experiment to test the 0.0 0.00
gravity model from last time. 0.1 0.25
0.2 0.75
Use a camera to record the position 0.3 1.50
every tenth of a second. 0.4 2.50
Data would be similar to the table; 0.5 4.00
plotted in the scatterplot below. 0.6 5.75
Distance !x"
Position of a dropped golf ball 0.7 7.75
15
0.8 10.25
0.9 13.00
10
1.0 16.00
Source: 5
practicalphysics.org [Ignore data on p. 25.]

Time !t"
0.0 0.2 0.4 0.6 0.8 1.0
Data Fitting — §1.3 & §3.2 13

Fitting Gravity Data

These data seem to fit a (type of function) . How can we be sure?

x as a function of t 2
1 Plot distance as a function of t 2. x

15

10

t2
0.2 0.4 0.6 0.8 1.0

2 ! Plot the log of distance as a function of log of time. !

log!x" as a function of log!t"
! Suppose x = Ct 2 . log of Distance

1.0

! Taking the log or ln of both sides gives 0.5

log x = log(Ct 2 ) = !1.0 !0.8 !0.6 !0.4 !0.2

log of Time

!0.5

! So all you need to do is fit to a line

log x ≈ 2 log t + 1.2
and solve for C !
Data Fitting — §1.3 & §3.2 14

Functions You Should Recognize on Sight

8 100 1000
6 80 800
60 600
4
40 400
2 20 200
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
3.0 1.2 1.0
2.5 1.0
2.0 0.8 0.5
1.5 0.6
1.0 0.4 2 4 6 8 10
0.5 0.2 !0.5
2 4 6 8 10 2 4 6 8 10 !1.0
0.4
8000 2
6000 0.3
1
4000 0.2
2000 0.1 2 4 6 8 10
!1
2 4 6 8 10 2 4 6 8 10
Data Fitting — §1.3 & §3.2 15

Fitting Gravity Data

We have determined that our gravity model x as a function of t
Distance !x"
x(t) = 16t 2 15

appears to model the dropping of a golf ball. 10

Is our model descriptively realistic? 5

Why? 0.0 0.2 0.4 0.6 0.8 1.0

Time !t"

Example. Raindrops—Our model gives their position as x(t) = 16t 2 .

A raindrop falling from 1024 feet would land after t = 8 seconds.
However, an experiment shows that the fastest drop takes 40 seconds,
and that drops fall at different rates depending on their size.

Even if we have a good model for one situation doesn’t mean it will
apply everywhere. We always need to question our assumptions.
—Extensive discussion in Section 1.3.—