Linear Models and Trendlines
Download
Report
Transcript Linear Models and Trendlines
Welcome to LSP 120
Dr. Curt M. White
What is LSP 120?
First course in QRTL (Quantitative
Reasoning and Technological Literacy)
Also known as Quantitative Reasoning
First Year Program requirement
Prereq to course is ISP 110 or Math 101
or placement through advising
Know this stuff already? Take the exam
this week.
What is LSP 120?
We will examine data – look at it, talk
about it, graph it, manipulate it
To graph and manipulate we will use
mathematics and technology
Why do we want to do this?
What is LSP 120?
Because the person that can use data
(information) gains knowledge and thus
has power!
In college – writing research papers and
taking advanced courses
In your job – career advancements
In government – getting things done (?)
In life – more intelligent life decisions
What is LSP 120?
“Information is a beacon, a cudgel, an
olive branch, a deterrent – all depending
on who wields it and how. Information is
so powerful that the assumption of
information, even if the information does
not actually exist, can have a sobering
effect.”
What are the topics?
Linear and near-linear models (with
trendlines and forecasting)
Exponential models
Useful percentages
Graphing
Consumer Price Index
Absolute versus relative values
Financial models
What is LSP 120?
Let’s take a closer look at the syllabus
LSP 120
Linear Models
and Trendlines
Linear Relationships
What makes a graph or function or table of
values linear? (You have already seen this!)
For a fixed change in x, there is a fixed
change in y, or
The change in y per unit change in x is
constant, or
There is a constant rate of change
Linear Relationships
Why do we care if a set of data is linear?
If a data set is linear (or near-linear), then
we can better predict where the data will
be in the future
We can also go back in time and see
where the data has been!
Linear Relationships
Look at a set of values. Is it a linear
relationship? Apply (B3-B2)/(A3-A2)
Column A Column B
(X values) (Y values)
2
3
4
5
6
7
4
5
6
7
8
9
25
50
75
100
125
150
No entry in this first row
=(B3-B2)/(A3-A2) = (50-25)/(5-4) = 25
=(B4-B3)/(A4-A3) = (75-50)/(6-5) = 25
=(B5-B4)/(A5-A4) = (100-75)/(7-6) = 25
=(B6-B5)/(A6-A5) = (125-100)/(8-7) = 25
=(B7-B6)/(A7-A6) = (150-125)/(9-8) = 25
All the results are the same (25), so this is a linear set of values.
Linear Relationships
If it is linear, what is the function?
Recall: y = mx + b
m is the slope, or the (change in y) / (change in
x)
b is the y intercept
So calculate the slope
Then plug in slope and first x and y values
into y=mx+b and solve for b
Linear Relationships
2
3
4
5
6
7
A(or X)
4
5
6
7
8
9
B(or Y)
25
50
75
100
125
150
No calculation in this row
(B3-B2)/(A3-A2) = (50-25)/(5-4) = 25
(B4-B3)/(A4-A3) = (75-50)/(6-5) = 25
(B5-B4)/(A5-A4) = (100-75)/(7-6) = 25
(B6-B5)/(A6-A5) = (125-100)/(8-7) = 25
(B7-B6)/(A7-A6) = (150-125)/(9-8) = 25
y = mx + b
m = change in y / change in x = 25/1 = 25
25 = 25 * 4 + b
b = -75 (this is the y-intercept)
y = 25x – 75
This is the equation of the line
Examples
x
5
10
15
20
y
-4
-1
2
5
x
0
2
4
6
y
1
4
16
36
m=change y / change x = 3/5=0.6
y = mx+b -4=0.6*5+b b=-7
y = 0.6x - 7
x
3
6
9
12
y
5
9
13
17
Rate of Change
=(B3-B2)/(A3-A2)
x
3
4
6
9
y
5
9
11
17
Linear Relationships
Linear growth – occurs when a quantity
grows by the same absolute amount
Exponential growth – occurs when a
quantity grows by the same relative
amount – that is by the same percentage –
in each unit of time
There is also linear decay and exponential
decay
Examples
The number of students at Wilson High School
has increased by 50 in each of the past four
years.
The price of milk has been rising with inflation at
3% per year.
Tax law allows you to depreciate the value of
your equipment by $200 per year.
The memory capacity of computer hard drives is
doubling approximately every two years.
The price of DVDs has been falling by about
25% per year.
Trendlines
Real data is seldom perfectly linear
Suppose the data is reasonably linear or
near-linear?
How does one make a linear model of the
data?
The standard approach is to use a “bestfit” line
In Excel – note the equation for the trendline and the R2 value.
Trendlines
If R2 = 1, then 100% of the variance in y is
explained by the line, so we have a perfect
fit (the data is linear)
If R2 = 0, then we have a terrible fit.
(Better not make a prediction)
What if R2 value = 0.5?
Trendlines
Do we have to do this calculation?
No, Excel can do it for you
Let’s say you have made a graph of some
data using Excel
After you make
a graph,
right-click on
any datapoint.
Select
Add Trendline…
Then check the
two bottom boxes
Trendlines
Let’s take a look at the dataset
Motorcycles_By_Year2005.xls
Graph the data using an XY Scatter. After
the graph is done, right click on a data
point, select Add Trendline
Trendlines
How do you make a prediction?
Two possible techniques:
1. Extend the trendline using Excel
2. Use the slope / intercept model (next slide)
Be careful! One’s confidence in
predictions made far from the data must
be tempered!!
Trendlines
You can use the equation for the trendline,
but the terms in the trendline equation are
significantly rounded.
Can remove some of this rounding
Right-click on the formula in the graph and
then click on Format Trendline Label
Select Number on left and right
Let’s give it 6 decimal places
Trendlines
You can also have Excel calculate the
Slope and Intercept individually, and then
use them in an equation
Somewhere in Excel, enter label Slope
and in next cell to right, enter =Slope(
Do the same thing for Intercept
Let’s Go!
Let’s head to the lab and start our first
Activity
But first, let’s break up into groups