Chip 1997 September/CHIP_CD_1997_09_PL.iso/software/mat_stat/statgraf/disk5/tutorial.txt

Text File | 1996-08-07 | 24.0 KB | 703 lines

WELCOME TO THE STATGRAPHICS PLUS FOR WINDOWS VERSION 2 DEMO

STATGRAPHICS Plus is an easy-to-learn, easy-to-use PC-based
statistics software package with an intuitive design that
will make your everyday data analysis easier than you ever
imagined.

This program is a working version of STATGRAPHICS Plus
Version 2 that offers features of the base system, Quality
Control, Design of Experiments, Time-Series Analysis,
Advanced Multivariate Methods, and Advanced Regression
Models modules. It will only analyze data files with a
maximum of 200 rows and 20 columns and the following
features are NOT operational:

        Saving
        Printing
        Changing text fonts
        Combining StatFolios or data files
        OLE
        Drag and drop

Use this demo to run analyses on your own data and to create
customized graphs. Or use our examples for a quick tour of
the program.

GETTING AROUND STATGRAPHICS PLUS

When you enter STATGRAPHICS Plus, it displays an untitled
data icon, a StatAdvisor icon, a StatGallery icon, and an
untitled Comments icon. When you are ready to start working,
you can either enter new data or use an existing data file
or StatFolio. Whichever method you chose, you'll be able to
explore STATGRAPHICS Plus Version 2 at your own pace.

EXAMPLE 1: SIMPLE REGRESSION AND STATFOLIO

In this example you will set up a Simple Regression and
become familiar with StatFolio, interactive graphics,
and StatAdvisor.

To Open a Data File:

1. Select File...Open Data File from the menu bar or click
the Open Data File button on the system toolbar.

Helpful Hint: For a description of the buttons on the
system and analysis toolbars, simply place the cursor
on the button and a description of the button appears
just below the cursor.

2. Double-click on the Data file/folder. Double-click
on cardata.sf from the scrolling list that appears below
the File Name text box. The system opens the data file
and displays the data icon with the file name.

To Perform a Simple Regression:

1. Select Relate...Simple Regression from the menu bar. The
system displays the Simple Regression data input dialog box.

2. Double-click on mpg.

3. Click on weight, then click the X arrow.

4. Click the OK button to display the Analysis Summary.

5. Click the Tabular Options button on the analysis toolbar
to display the Tabular Options dialog box. Select one or
as many options as you like. To follow our example, select
Analysis Summary, Forecasts, and Comparison of Alternative
Models.

6. Click the OK button.

Helpful Hint: To see other options that are available for an
analysis, click the right mouse button on the pane of that
analysis and select Analysis Options...

To maximize a pane, double-click on it; to minimize it,
double-click again.

7. Click the Graphics Options button on the analysis
toolbar. The system displays the Graphical Options dialog
box. Select one or as many options as you like. To follow
our example, select Plot of Fitted Model and Observed
versus Predicted.

8. Click the OK button.

9. After reviewing the results, click the Minimize button
(the down arrow in the upper right corner of the analysis
window) to shrink the window to an icon.

Helpful Hint: Example 1 and all of the following examples
have been saved in StatFolios for you. If you would like to
leave the program, you may do so at any time. If you want to
continue working with an example when you restart the
program, select File...Open StatFolio from the menu bar.
Double-click on the Data file/folder. Then select the
appropriate example*.sgp file.

Improving Your Productivity with StatFolios:

Imagine that you perform a specific series of analyses each
year on the mileage and weight of a sample of cars. Each
year you store the data in a new file; each year's data
file contains the same variables. To save time, you can
use StatFolios to automatically update the analyses to
reflect the new data.

1. Continue from Example 1 or select File...Open StatFolio
from the menu bar.

2. Double-click on example1.sgp from the scrolling list that
appears below the text box.

3. Double-click the Simple Regression icon to maximize the
Simple Regression Analysis from Example 1.

To visually compare the differences in the output that the
two datasets produce, maximize one of the graphs.

4. Select File...Open Data File from the menu bar or click
the Open Data File button on the system toolbar.

5. Double-click on cardata2.sf from the scrolling list that
appears below the text box. The system opens the new data
file and automatically updates all your analyses with the
new data.

You can also change the variables in the analysis.

1. Click the Input dialog button on the analysis toolbar.
The system displays the Simple Regression data input dialog
box.

2. Click on horsepower, then click the X arrow.

3. Click the OK button to automatically update your analysis
with mpg and horsepower.

Interactive Graphics at Your Fingertips:

Within every analysis STATGRAPHICS Plus gives you
spectacular graphics that let you interact with your data.
Your graphics become an analytical tool, not just a
presentation device. You can quickly and easily explore
your data, discovering new relationships, while the graph
is displayed on the screen.

1. Continue from Example 1 or select File...Open StatFolio
from the menu bar.

2. Double-click on example1.sgp from the scrolling list that
appears below the text box.

3. Double-click the Simple Regression icon to maximize the
Simple Regression analysis from Example 1.

4. Double-click on Plot of mpg to maximize the pane.

To Brush Points on a Graph:

Brushing is a technique that allows you to identify data
points on a graph in a different color to examine the
relationship of an additional variable on the existing
variable(s).

1. Click the Brush button on the analysis toolbar to display
the Brushing dialog box.

2. Double-click on horsepower.

3. Click the OK button. The system displays a slide bar on
the analysis toolbar.

4. Move the slide bar to the right to examine the
relationship of horsepower on mpg as horsepower increases.
Move the slide bar to the left to examine the relationship
of horsepower on mpg as horsepower decreases.

To Display a Label for a Specified Point:

1. Click the Identify button on the analysis toolbar. The
system displays the Point Identification dialog box.

2. Double-click on make.

3. Click the OK button.

4. Click on a point in the graph. The system displays the
corresponding label in the Label text box and the corresponding
row number in the Row text box.

Using StatAdvisor:

StatAdvisor is an exclusive feature of STATGRAPHICS Plus. It is
a tool that allows you to display and print a short,
easy-to-understand interpretation of an analysis.

StatAdvisor is available for use with most statistical analyses.
The interpretation includes a simple explanation of the results,
which are data sensitive; that is, the interpretation varies
according to the variables you use in the analysis. For example,
when you use StatAdvisor after you perform a Multiple Regression
analysis, the interpretation indicates which variables appear to
add significantly to the fitted model.

You can access StatAdvisor by:

  - clicking the StatAdvisor icon on the application window
  - clicking the StatAdvisor button on the application toolbar
  - selecting Help...The StatAdvisor... from the menu bar

EXAMPLE 2: X-BAR AND R CHART AND STATGALLERY

In this example you will set up an X-bar and R chart and learn
how to use StatGallery. StatGallery is an area you use to place
one or multiple text and graphics panes in a variety of positions
so you can view or print them.

To Open a Data File:

1. Select File...Open Data File from the menu bar or click the
Open Data File button on the system toolbar.

2. Double-click on qcdata.sf from the scrolling list that appears
below the File Name text box. The system opens the data file and
displays the data icon with the file name.

To Perform an X-bar and R Chart:

1. Select Special...Quality Control...Variables Control Charts...
X-bar and R from the menu bar. The system displays the X-bar
and R Charts data input dialog box.

2. Double click on cereal.

3. Click on the Subgroup Numbers or Size text box and type 5.

4. Click the OK button. The system displays the initial
analysis window.

5. Click the Tabular Options button on the analysis toolbar to
display the Tabular Options dialog box. To follow our example,
select All.

6. Click the OK button.

7. Click the Graphics Options button on the analysis toolbar to
display the Graphical Options dialog box. Select one or as many
options as you like. To follow our example, select X-bar Chart,
 Range Chart, and Tolerance Chart.

8. Click the OK button.

Helpful Hint: To exclude a subgroup from an analysis, double-click
the X-bar Chart to maximize the window. Click the point that is
below the lower control limit (Subgroup 20). Click the
Exclude/Include button on the analysis toolbar.

The system excludes the point from the analysis, replots the
charts, and recalculates all of the tabular options.

Using StatGallery:

1. Continuing from Example 2, click the right mouse button on the
X-bar Chart.

2. Select Copy to Gallery on the pop-up menu. The Copy to Gallery
Options dialog box displays. The default StatGallery arrangement
is two by two.

3. Select Top Left and click the OK button. The program moves the
X-bar Chart to the StatGallery window.

4. Minimize the X-bar Chart.

5. Select another item (text or graph). Click the right mouse
button and select Copy to Gallery on the pop-up menu.

6. The system automatically highlights the next available
position. Click the OK button.

7. To view the items in StatGallery, double click on the
StatGallery icon.

To Modify the Arrangement:

STATGRAPHICS Plus allows you to rearrange the position
of the items in StatGallery.

1. Click the right mouse button while in the StatGallery window.

2. Select Modify Arrangement...on the pop-up menu. The
StatGallery Options dialog box displays.

3. Select Top and Bottom.

4. Click the OK button.

EXAMPLE 3: CREATING AND ANALYZING A SCREENING DESIGN

In this example you will learn how to create a screening
design within Design of Experiments.

1. Select Special...Experimental Design...Create Design
from the menu bar. The system displays the Create Design Options
dialog box.

2. Select Screening under the Design Class option;
accept the default in the No. of Response Variables text box;
type 4 in the No. of Experimental Factors text box; and
type Airplane Experiment in the Comment text box.

3. Click the OK button; the system displays the Factor
Definition Options dialog box.

4. To complete the Factor Definition dialog box type Design
in the Name text box; type Straight in the Low text box;
type Tee in the High text box; deselect the
Continuous check box.

Helpful Hint: Use the Tab button to move between text boxes.

Click the B Factor button. For Factor B, type Nose in the
Name text box; type None in the Low text box; type
Clip in the High text box; deselect the Continuous check box.

Click the C Factor button. For Factor C, type Paper in the
Name text box; type Notebook in the Low text box; type
Construct in the High text box; deselect the Continuous
check box.

Click the D Factor button. For Factor D, type Wing in the
Name text box; type Straight in the Low text box; type Bent
in the High text box; deselect the Continuous check box.

5. Click the OK button; the Response Definition Options
dialog box displays.

6. Type Distance in the Name text box; type Feet in the
Units text box.

7. Click the OK button; the system displays the Screening
Design Selection dialog box.

8. Use the Down Arrow button to the right of the dropdown
single-selection list to display the available designs.
Select Factorial 2^4.

9. Click the OK button; the system displays the Screening
Design Options dialog box.

10. Deselect the Randomize check box so the system will
create the design in a standardized order to make it easier
for you to enter data.

11. Click the OK button; the system displays the Screening
Design Attributes window.

Minimize the Screening Design Attributes window. Now
you're ready to collect and set up the data for the
experiment.

Entering the Experimental Results:

1. Double-click the untitled data icon. The system displays
the spreadsheet.

2. Tab to the Distance column and enter the following values:

    6.25, 15.5, 7.0, 16.5, 4.75, 5.5, 4.5, 6.0,
    7.0, 10.0, 10.0, 16.0, 4.5, 6.0, 4.5, 5.75

Helpful Hints: Remember, Example 3 has been saved in a StatFolio
for you. If you would like to leave the program, you may do so
at any time. If you want to continue working with this example
when you restart the program, select File...Open StatFolio
from the menu bar. Then select the example3.sgp file.

Data files have an .sfx extension in the Design of Experiments
module.

Analyzing the Experimental Results:

1. Continue from Example 3.

2. Select Special...Experimental Design...Analyze Design
from the menu bar. The system displays the Analyze Design
data input dialog box.

3. Double-click on Distance.

4. Click the OK button. The system displays the Analysis Summary.

5. Click the Graphics Options button on the analysis toolbar
to display the Graphical Options dialog box. Select one or
as many options as you like. To follow our example, select
Pareto Charts.

6. Click the OK button. The Pareto Chart helps determine
which factors contribute significantly to flight distance.

7. Click the Tabular Options button on the analysis toolbar
to display the Tabular Options dialog box. Select one or as
many options as you like. To follow our example, select
ANOVA Table.

8. Click the OK button. The ANOVA Table shows that Factors A, C,
and AC are statistically significant.

EXAMPLE 4: USING SEASONAL DECOMPOSITION

In this example you will analyze monthly sales of a major U.S.
automobile manufacturer over a five-year period.

To Open a Data File:

1.  Select File...Open Data File from the menu bar or
click the Open Data File button on the system toolbar. The
system displays the Open Data File dialog box.

2. Double-click on carsales.sf from the scrolling list that
appears below the File Name text box. The system opens the
data file and displays the data icon with the file name.

To Perform a Seasonal Decomposition:

1. Select Special...Time-Series Analysis...Seasonal Decomposition
from the menu bar. The system displays the Seasonal Decomposition
data input dialog box.

2. Double-click on sales.

3. Click the Month(s) option button.

4. Highlight 1/50 in the Starting At text box and type 1/82.
This indicates that the data you want to use begins in
January 1982.

5. Click the Seasonality text box and type 12. Accept the
defaults for the other text boxes.

6. Click the OK button.

7. Click the Tabular Options button on the analysis toolbar to
display the Tabular Options dialog box. Select one or as many
options as you like. To follow our example, select Analysis
Summary and Data Table.

8. Click the OK button.

9. Click the Graphics Options button on the analysis toolbar.
The system displays the Graphical Options dialog box. Select
one or as many options as you like. To follow our example,
select Trend-Cycle, Seasonal Indices, and Seasonally Adjusted
Data.

10. Click the OK button.

EXAMPLE 5: CLUSTER AND DISCRIMINANT ANALYSIS

In this example you will investigate 16 growth funds to determine
how well the funds fared. Using Cluster Analysis you will see
how the funds group and determine the characteristics of the
clustered groups.

Using Discriminant Analysis, you will use the risk, expense, tax
ratings, and performance over a five-year peiod to discriminate
between recommendations. If the discriminant analysis results are
significant, you can then classify new funds.

To Open a Data File:

1. Select File...Open Data File from the menu bar or click the
Open Data File button on the system toolbar. The system displays
the Open Data File dialog box.

2. Double-click on growth.sf from the scrolling list that appears
below the File Name text box. The system opens the data file and
displays the data icon with the file name.

To Perform a Cluster Analysis:

1. Select Special...Multivariate Methods...Cluster Analysis from
the menu bar. The system displays the Cluster Analysis data input
dialog box.

2. Double-click on Expense, Five_Yr, Perf90, Perf91, Perf92,
Perf93, Perf94, Risk, and Tax.

3. Enter Fund into the Point Labels text box.

4. Click the OK button. The system displays the Analysis Summary.

5. Click the right mouse button on the Analysis Summary.
Select Analysis Options to display the Cluster Analysis
Options dialog box.

6. Click the Ward's option button and accept the defaults in the
other boxes. Because you want to use a measurement method that
combines clusters that have a same number of observations and
that produces approximately the same number of observations,
Ward's method of clustering is used.

7. Click the OK button; the Analysis Summary redisplays.

8. Click the Graphics Options button on the analysis toolbar. The
system displays the Graphical Options dialog box. Select one or
as many options as you like. To follow our example, select
Dendrogram.

9. Click the OK button. The dendrogram shows the mutual funds
as they are clustered. There seems to be at least three distinct
groups.

10. Click the right mouse button on the Analysis Summary.
Select Analysis Options to display the Cluster Analysis
Options dialog box.

11. Change the Number of Clusters from 1 to 3.

12. Click the OK button.

13. Click the Tabular Options button on the analysis toolbar
to display the Tabular Options dialog box. Select one or as
many options as you like. To follow our example, select
Membership Table and Icicle Plot.

14. Click the OK button.

To Perform a Discriminant Analysis:

1. Continue from Example 5.

2. Select Special...Multivariate Methods...Discriminant Analysis
from the menu bar. The system displays the Discriminant Analysis
data input dialog box.

3. Double-click Recommend.

4. Enter Expense, Five_Yr, Risk, and Tax into the Data text box.

5. Enter Fund into the Point Labels text box.

6. Click the OK button; the system displays the Analysis Summary.

7. Look at the summary; scroll to the bottom of the summary.
StatAdvisor indicates the p-values are statistically significant.

8. Click the Tabular Options button on the analysis toolbar.
The system displays the Tabular Options dialog box. Select one
or as many options as you like. To follow our example, select
Classification Functions and Classification Table.

9. Click the OK button.

10. Click the Graphics Options button on the analysis toolbar.
The system displays the Graphical Options dialog box. Select
one or as many options as you like. To follow our example, select
2D Scatterplot.

EXAMPLE 6: FITTING A CALIBRATION LINE

Calibration uses the regression model Y = aX + b to predict
values for X as well as for Y. In this example you will use
the Calibration Models Analysis to calibrate the test method
by quantifying the relationship between absorbancy (y) and
concentration (x).

To Open a Data File:

1. Select File...Open Data File from the menu bar or click the
Open Data File button on the system toolbar. The system displays
the Open Data File dialog box.

Helpful Hint: If you installed STATGRAPHICS from the CD or
a floppy disk set, please make sure you copied calib.sf and
example6.sgp to your sgwin\data directory or open the file
directly from your CD or floppy disk drive.

2. Double-click on calib.sf from the scrolling list that appears
below the File Name text box. The system opens the data file and
displays the data icon with the file name.

Determining the Relationship Between Two Variables:

1. Select Special...Advanced Regression...Calibration Models...
from the menu bar. The system displays the Calibration Models
dialog box.

2. Click the absorbance variable, then click the Y (Measured)
text box arrow.

3. Click the concent variable, then click the X (Actual)
text box arrow.

4. Click the OK button. The system displays the Analysis Summary.

5. Click the right mouse button on the Analysis Summary.
Select Analysis Options to display the Calibration Model
Options dialog box.

6. Click the check box for Include Constant to turn it off,
then click the OK button. The program recalculates and
redisplays the Analysis Summary.

7. Click the Graphics Options button on the analysis toolbar. The
system displays the Graphical Options dialog box. Select one or
as many options as you like. To follow our example, select
Plot of Fitted Model.

Helpful Hint: The example6.sgp StatFolio has been saved for you up
until this point.

9. Click the OK button. Notice that because you removed the
constant, the prediction limits are not quite parallel;
instead, they are slightly closer together for lower values
of the concentration. This is more evident for the confidence
limits, so we will add them to the plot.

10. Click the right mouse button on the graphics pane.
Select Pane Options to display the Plot of Fitted Model
Options dialog box.

11. Click on Confidence Limits, accept the defaults
for the other options.

12. Click the OK button. The program recalculates the
redisplays the plot with the confidence limits.

Calculating a Prediction for a New Sample:

1. Click the right mouse button on the graphics pane.
Select Pane Options to display the Plot of Fitted Model
Options dialog box.

2. Click on Confidence Limits to turn it off, click
the X Predict option, type 0.352 in the At text box,
accept the defaults for the other options.

3. Click the OK button; the program plots the prediction
lines and the concentration values and redisplays the plot.

Creating a Predictions Table:

1. Click the Tabular Options button on the analysis toolbar
to display the Tabular Options dialog box. Select one or as
many options as you like. To follow our example,
select Predictions.

2. Click the OK button. Notice the table lists predictions
for Y for the high and low values of X. You can use the
regression equation to predict either X or Y, given a value
for the other variable.

Predicting X Values for Six Different Y Values:

1. Click the right mouse button on the Predictions pane.
Select Pane Options to display the Predictions Options
dialog box.

2. Click the X Predict option; enter 0, 0.1, 0.2, 0.3, 0.4, 0.5
in the first six Predict at: text boxes. Accept the defaults
for the other options.

3. Click the OK button. The table shows predictions
for the concentrations of six absorbency readings as
well as the prediction intervals.

THE RIGHT SOLUTION FOR YOU

STATGRAPHICS Plus will make your everyday data analysis easier
than you ever imagined. StatFolios will save you time and the
aggravation of redoing routine analyses. And our unique
interactive graphics will let you explore your data and
discover new relations while the graphs are on your screen.
Our new StatAdvisor lets you display and print short,
easy-to-understand interpretations of an analysis and
StatGallery lets you place one or multiple text and graphics
panes in a variety of positions so you can view or print
them. We're sure you'll find everything you've been looking
for in a statistics package in STATGRAPHICS Plus.

FOR MORE INFORMATION

We hope you experienced a new dimension in data analysis with
STATGRAPHICS Plus!

To order your copy of STATGRAPHICS Plus Version 2, call:

800-592-0050 in the United States
301-984-5412 Outside the United States
301-984-5094 Telefax
www.statgraphics.com

For a listing of international STATGRAPHICS dealers, please
refer to the DEALER.TXT file on this CD.