You are on page 1of 2

120MP Lab 8 Answers Regression Analysis in Minitab

The following data regarding a particular model of car are in the first sheet of the spreadsheet
Lab17.xls

Mileage (thousands) 19 41 33 59 96 67
Price (£) 3250 2650 2100 1650 1250 1100

Copy and Paste the data into Minitab.

Familiarise yourself with the following Minitab commands and output for carrying out a
regression analysis.

Click on Stat > Regression > Fitted Line Plot


Select the variables
Click Graphs > Residuals versus Fits > OK
Options > Display confidence interval and Display prediction interval > OK
Storage > Fits > OK > OK

This will give you;


A scatter diagram with the regression line and confidence / prediction bands,
the equation of the regression line,
the standard error of the fit s,
the coefficient of determination R2,
a plot of the residuals,
the fitted y values y ,
the Total sum of squares,
the Residual sum of squares,

You could obtain the correlation coefficient r by simply square rooting R-square on a
calculator, or using
Stat > Basic Statistics > Correlation in Minitab

You can obtain predictions for other values of x using


Stat > Regression > Regression > Predict

Select relevant output, add comments, and save as follows:


Click on to Show Session Folder
Right-click on relevant output item and Append to Report
Click on to Show Report Pad and add comments explaining and interpreting the output
File > Save Report As

120MP 1
Trees Exercise
The second sheet of the Lab17.xls spreadsheet contains the trunk diameter and height (feet),
and the trunk volume (cubic feet) for 31 cherry trees.
The purpose of this exercise is to find a way of predicting the volume of wood in a tree from
either its diameter, height, or some simple combination of both measurements.

1. Copy the data into Minitab.


2. Plot the variables against each other on separate graphs. What relationships do you see?
Strong correlation between volume and diameter (0.967)

3. What do you notice about the relationship between volume and height?
Variability in volume increases with height

4. Show that diameter is a better predictor of volume than height? Why is this do you think?
Correlation between volume and height is only 0.598
Height of tree is not the same as length of trunk – could have long branches

5. What is the regression equation for predicting volume from diameter? What value is R-Sq?

Volume = -36.96 +60.8 Diameter Rsq = 93.5%


6. Plot the residuals against the diameter. What do you conclude?
Hint of non-linearity. Tree 31 has a large positive residual.

7. Obtain a (point) prediction for the volume of a cherry tree of diameter 9 inches. Also give a
95% prediction interval, and comment on it
8.65 (-0.44 , 17.73) very wide and includes zero

8. Assuming that a tree is roughly cylindrical, what combination of height and diameter do you
think would be a better predictor of a tree’s volume?
H x D2

9. Use the Calc > Calculator facility to create this combination as a new variable.
10. Using this new combination as your predictor variable, find a new regression line to predict
the volume of a cherry tree, and investigate how good a fit it is
V = -0.297 + 0.306HD2 R2 = 97.8% residual plot random
So good fit
11. Save your report / project

120MP 2

You might also like