Yournotes sponsored in part by

Study Break!

Regression Regression

Lecture Goals

 

You should be able to:

    - calculate the least squares regression line and use it to predict new values

    - understand and calculate the standard error of the estimate

     
Reading

Chapter 7

 

We find a positive relationship between study time and exam scores.

Now we want to predict how much time we need to study to earn an A.

How predict exam score based on study time???

 

Student     Study time     Exam score

    A                 8                    88

    B                 4                     75

    C                 8                     85

    D                 5                     78

    E               10                     94

    F               12                   100

    G                 6                     82

    H                 3                     71

 
How predict exam score based on study time?

 
Could get average scores...
 
 

Study Time     Average Exam Score

        3                     71.5

        4                     75.5

        5                     75             What about

        6                     82.5         studying for

        8                     87.3         7 or 11 hrs.?

      10                     94.3

      12                     94.6

 

= use relationship between to variables for prediction

  = given # drinks can we predict # errors?

 

= given data can we predict dosages??

 

= based on Line of Best Fit idea

 

Line Of Best Fit

     Best representation = line that is closest to all coordinates

  1) Visual approximation

 

2) Use statistics to determine formula for line of best fit
    (regression line)

 
Characteristics of Lines

 

1) The line itself is made up of many (x,y) coordinates

(graph)

 

2) Lines have a slopes

(graph)

 

3) If we extend the line it will intersect the y-axis somewhere

(graph)

 

 

Formula for lines MUST capture these defining characteristics:

- Made up of (x, y) coordinates

- slope

- y-intercept

 

Formula for ANY Line

 

Y = bX + a

 

Y = y coordinate of (x, y) pair

X = x coordinate of (x, y) pair

 

b = slope of the line

a = y-intercept

 

Slope of the Line

 

= rate of climb or fall

 

b = change in Y values
       change in X values

 

b = Y1-Y2
      X1-X2

 

Any 2 sets of paired values are fine

(graph)

 

= rate of climb or fall

 

b = change in Y values
      change in X values

 

b = 8-10 = -2 = 1 = .10
      30-50  -20 10

 

(graph)

= rate of climb or fall

 

b = change in Y values
       change in X values

 

b = 8-9 = -1 = .10
     0-40  -10

 

Y-Intercept

 

= where line meets y-axis

= y-value when x = 0

  (graph)

 

Least Squares Regression Line

 
= line of best fit determined by calculation

 

= line closest to all points (least error)

 

= line that minimizes the sum of the squared errors to all points

  (graph)

 

(graph)

 
= used to predict the Y coordinate for any given X coordinate

= used to predict # errors given # drinks

 

Formula: Y' = bX + a

 
 

Slope of Regression Line

 
(formula)

 
 
Y-Intercept of Regression Line
 

a = (mean of Y) - b(mean of X)

Student     IQ     GPA

    1         110     1.0

    2         112     1.6

    3         118     1.2

    4         119     2.1

    5         122     2.6

    6         125     1.8

    7         127     2.6

    8         130     2.0

    9         132     3.2

  10         134     2.6

  11         136     3.0

  12         138     3.6

 

Q - Which is x and which is y?

 
A - Which one is given and which one do you want to predict?

 

If want to predict IQ then GPA=X IQ=Y

 

If want to predict GPA then GPA=Y IQ=X

   

In this case IQ = X and GPA = Y

Formula: Y' = bX + a

 

a = (mean of Y) - b(mean of X)

What do we need????

 

(symbols)

 

 

 

 

 

 

 

 

 

 

Student     IQ     GPA

    1         110     1.0

    2         112     1.6

    3         118     1.2

    4         119     2.1

    5         122     2.6

    6        

Information contained on this page does not represent the lecture verbatim.
These notes are not a substitute for class attendance.



This page last updated: [an error occurred while processing this directive]
Copyright 1998.
Questions?  Email: info@yournotes.com