SlideSlide 1

Warm Up: #23, 24 on page 555

Answer each of the following questions for #23, 24:a) Is there a linear correlation? Use software then Table A-6 to prove it. Are your

answers the same? b) Graph the points (don’t forget axis labels). If there is a correlation, graph the

LSRL and continue to do c-h.c) Find the vital statistics (r, r-squared, a, b, y-hat – don’t forget to define x and y)d) Tell me what r and r-squared means in the context of the problem (r: form,

direction, strength) (r-squared: how much of the variation in x can be explained by the variation in y)

e) Find the residualsf) Draw the residual plot – is the regression line a good model for the data? Why?g) For # 23, predict the winning time when the temperature is 73 degrees

Fahrenheit. h) For #24, predict the height of a daughter when her mother is 66 inches tall.

SlideSlide 2

The SAT essay: longer is better?

Words 460 422 402 365 357 278 236 201 168 156 133 114 108 100 403

Score 6 6 5 5 6 5 4 4 4 3 2 2 1 1 5

Words 401 388 320 258 236 189 128 67 697 387 355 337 325 272 150

Score 6 6 5 4 4 3 2 1 6 6 5 5 4 4 2

Words 135 73

Score 3 1

SlideSlide 3

Section 10-4 Variation

SlideSlide 4

Key Concept

In this section we proceed to consider a method for constructing a prediction interval, which is an interval estimate of a predicted value of y.

Using paired data (x,y), we describe the variation that can be explained between x and y and the variation that is unexplained.

SlideSlide 5

Figure 10-9

Unexplained, Explained, and Total Deviation

SlideSlide 6

DefinitionsTotal Deviation The total deviation of (x, y) is the vertical distance y – ybar, which is the distance between the point (x, y) and the horizontal line passing through the sample mean y-bar.Explained DeviationThe explained deviation is the vertical distance y-hat

- y-bar, which is the distance between the predicted y-value and the horizontal line passing through the sample mean y-bar.

Unexplained Deviation The unexplained deviation is the vertical distance y – y-hat, which is the vertical distance between the point (x, y) and the regression line. (The distance y – y-hat is also called a residual, as defined in Section 10-3.)

SlideSlide 7

Particulars

We can explain the discrepancy between y-bar = 9 and y-hat=13 by noting that there is a linear relationship best described by the LSRL (y-y-hat).

The discrepancy between y-hat = 13 and y=19 can’t be explained by the LSRL = residual or unexplained deviation (y-y-hat)

SlideSlide 8

(total deviation) = (explained deviation) + (unexplained deviation)

(y - y) = (y - y) + (y - y)^ ^

(total variation) = (explained variation) + (unexplained variation)

(y - y) 2

= (y - y) 2

+ (y - y) 2^ ^

Formula 10-4

Relationships

SlideSlide 9

Definition

r2 =explained variation.

total variation

The value of r2 is the proportion of the variation in y that is explained by the linear relationship between x and y.

Coefficient of determinationis the amount of the variation in y thatis explained by the regression line.

SlideSlide 10

Warm Up: Day 2

Consider the following data set:

Find:

a) Total variation

b) Explained variation

c) Unexplained variation

X Y

1 4

2 24

4 8

5 32

SlideSlide 11

Try again!

Consider the following data set:

Find:a) Total variationb) Explained variationc) Unexplained variation

X Y

1 1

2 3

3 5

4 7

SlideSlide 12

Not Old Faithful again!

In section 10-2 we used the duration/interval after eruption times in Table 10-1 to find that r = .926. find the coefficient of determination. Also, find the percentage of the total variation in y (time interval after eruption) that can be explained by the linear relationship between the duration of time and the time interval after an eruption.

Duration 240 120 178 234 235 269 255 220

Interval After 92 65 72 94 83 94 101 87

SlideSlide 13

Interpretation/New Def86% of the total variation in time intervals after eruptions (y) can

be explained by the duration times (x)14% of the total variation in time intervals after eruptions can be

explained by factors other than duration times.

Recall: y-hat = 34.8 +.234x (x = duration in seconds, y = predicted time interval). When x = 180, we predict a y-hat of ____?

This single value is called a point estimate. It is our best predicted value. How accurate is it?

We use prediction intervals to answer this question.

SlideSlide 14

DefinitionsPrediction Interval: an interval estimate of a predicted value

of y. The development of a prediction interval requires a measure of the spread of sample points about the regression line.

The standard error of estimate, denoted by se is a measure of the differences (or distances) between the observed sample y-values and the predicted values y that are obtained using the regression equation. That is, it is a collective measure of the spread of the sample points about the regression line.

Se = A measure of how sample points deviate from their regression line.

SlideSlide 15

Standard Error of Estimate

se =

or

se = y2 – b0 y – b1 xyn – 2 Formula 10-5

(y – y)2

n – 2

^

SlideSlide 16

Given the sample data in Table 10-1, find the standard error of estimate se for the duration/interval data.

Example: Old Faithful

Duration 240 120 178 234 235 269 255 220Interval After 92 65 72 94 83 94 101 87

SlideSlide 17

y - E < y < y + E^ ^

Prediction Interval for an Individual y

where

E = t2 se n(x2) – (x)2

n(x0 – x)2

1 + +1n

x0 represents the given value of x

t2 has n – 2 degrees of freedom

SlideSlide 18

E = t2 se +

n(x2) – (x)2

n(x0 – x)2

1 + 1

n

Example: Old FaithfulFor the paired duration/interval after eruption times in Table 10-1, we have found that for a duration of 180 sec, the best predicted time interval after the eruption is 76.9 min. Construct a 95% prediction interval for the time interval after the eruption, given that the duration of the eruption is 180 sec (so that x = 180).

Duration 240 120 178 234 235 269 255 220Interval After 92 65 72 94 83 94 101 87

SlideSlide 19

y – E < y < y + E

76.9 – 13.4 < y < 76.9 + 13.4

63.5 < y < 90.3

^ ^

Example: Old Faithful - contFor the paired duration/interval after eruption times in Table 10-1, we have found that for a duration of 180 sec, the best predicted time interval after the eruption is 76.9 min. Construct a 95% prediction interval for the time interval after the eruption, given that the duration of the eruption is 180 sec (so that x = 180).

SlideSlide 20

Same problem, different x

For the paired duration/interval after eruption times, find:

1) For a duration of 150 sec, the best predicted time interval after the eruption is _____ min.

2) Construct a 95% prediction interval for the time interval after the eruption, given that the duration of the eruption is 150 sec (so that x = 150).

Duration 240 120 178 234 235 269 255 220

Interval After 92 65 72 94 83 94 101 87

E = t2 se +

n(x2) – (x)2

n(x0 – x)2

1 + 1

n

y – E < y < y + E^ ^