文档库

最新最全的文档下载
当前位置:文档库 > Design Expert说明书

Design Expert说明书

Stat-Ease, Inc. offers classes on design of experiments (DOE) for everyone from the first-time experimenter to the advanced practitioner. Whether you are an engineer, scientist, quality professional or statistician, you’ll want to do all you can to improve quality and efficiency, and save money at the same time!

By enrolling in a Stat-Ease?class you will benefit from experienced instructors (see http://www.wendangku.net/doc/a375b4c45fbfc77da269b19a.html for pictures and biographies) who have worked in industry. You will also receive free follow-up help by phone, fax, or e-mail. You’ll work through many hands-on case study exercises to learn how to apply your new skills. Workshops designated with a icon are computer intensive.

Students attending these classes will be given a path to all simulation and data files used in class, which are posted to a special Internet site where they can also link to a free, fully-functional, but time-limited, copy of Design-Expert software for use after class.

In addition to public workshops in the United States, Stat-Ease offers private in-house classes at your location, both here and abroad. In-house training becomes cost-effective when you have six or more students. For more information on any of our workshops, please contact Elicia at 612.746.2038. Visit our web site at http://www.wendangku.net/doc/a375b4c45fbfc77da269b19a.html.

COURSE DESCRIPTIONS

DOE Simplified—An Intro to Breakthrough T ools for Planned Experimentation (DOES) - 1 day In-House Only

This one-day workshop offers the perfect introduction to DOE for technical professionals and their managers. DOE Simplified provides an overview of DOE principles and illustrates an array of tools and design types. In the morning you are introduced to the concepts of DOE and then explore full factorials, interactions and aliasing. Several fun and informative case studies from the book* are presented, such as making microwave popcorn and starting a reluctant weed whacker. The afternoon session looks at how to apply general factorials, response surface designs and mixture designs. DOE Simplified is presented in an easy-to-digest, light and humorous manner with fun case studies. Participants will learn to appreciate all that DOE can do to find their sweet spot—the ideal settings for process factors and/or recipes for mixture experiments. While not a substitute for hands-on computer training, this workshop provides an educational starting point. (If you would like to dive right into DOE, we suggest you go directly to the Experiment Design Made Easy workshop.)

Prerequisites:None. Fee:Private workshop only.Fee Includes:DOE Simplified, 2nd Edit ion text, with a 180-day, time-limited, yet fully functional, copy of Design-Ease?7.1 software (students may purchase the permanently-licensed version at 20% off) and course materials (.8 CEU’s).

Design Expert说明书

Statistics for T ech. Professionals

In-House Only

Learn how to extract information from data by attending the new computer-intensive

Design of Experiments (DOE) Workshops

Learn how to optimize your product or process with DOE Our objective is to keep it simple and make it fun. After

mastering these simple tools, you’ll be ready to move on

to the next level, the three-day Experiment Design Made

Easy workshop.

PreDOE is not an academic course—it’s a hands-on

presentation of practical statistical tools. You will learn

to use basic statistics to describe your data, do

hypothesis testing, and look at ANOVA concepts. You no

longer need to be intimidated by statistics. PreDOE

empowers you to use it to your benefit.

Upon successful completion of the course (which

involves passing the final test with a score of 80% or

higher) you'll receive a certificate granting you 0.6 CEU's.

or more information, please visit our website:

http://www.wendangku.net/doc/a375b4c45fbfc77da269b19a.html/out_pre.html.

Prerequisites:None. Determine whether or not you

should take this course with our free Self-Assessment

Questionnaire at http://www.wendangku.net/doc/a375b4c45fbfc77da269b19a.html. Fee:$95.00

(0.6 CEU’s).

Stat-Ease, Inc., Hennepin Square, Suite 480, 2021 E. Hennepin Ave.

Minneapolis, MN 55413, http://www.wendangku.net/doc/a375b4c45fbfc77da269b19a.html

p: 612.378.9449, f: 612.746.206929I0803“Stat-Ease courses enabled me to bring better products to market faster. They were truly among the best classes I have attended (including graduate-level classes) in terms of improving my abilities to succeed in product development.”

Alex Fensore International Imaging Materials (New York) Diagram of the Workshop Prerequisites

Shorter Courses

Design Expert说明书

DOE for DFSS (DDFSS)—2 days

Use DOE to create products and processes robust to Mixture Design for Optimal Formulations If you do product formulation, you need the mixture Response Surface Methods for Process Response Surface Methods (RSM) lead you to the peak of Experiment Design Made Easy (EDME)—3 days

Learn how to make breakthrough improvements using Three-Day Core Classes

DX71-02A-Gen1Factor-P1 Rev. 4/15/07 General One-Factor Tutorial

(Part 1 – The Basics)

Introduction

In this tutorial you will build a general one-factor design using Design-Expert? software.

This type of design can be very useful for simple comparisons of categorical treatments,

such as:

?Who will be the best supplier,

?Which type of raw material should be selected,

?What happens when you change procedures for processing paperwork.

If you wish to experiment on a continuous factor, such as time, which can be adjusted to

any numerical level, consider making use of response surface methods (RSM) instead.

This is covered in a separate tutorial.

The data for this example come from the Stat-Ease bowling league. Three bowlers must

compete for the last position on the team. They each bowl six games (see data below).

Game Pat Mark Shari

1 160 165 166

2 150 180 158

3 140 170 145

4 167 18

5 161

5 157 195 151

6 148 175 156

Mean 153.7 178.3 156.2

Bowling scores

The captain knows better than to just simply pick the bowler with the highest score.

Maybe it’s a fluke that Mark scored highest and Pat’s score is low. He wants to know if

the scores are significantly different, given the variability in individual scores.

This one factor case study provides a good introduction to the power of simple

comparative design of experiments (DOE). It will exercise a number of handy features

provided by Design-Expert software. We won’t explain all features displayed – some

will be covered in follow-up tutorials. Many other features and outputs will be covered

only in the help system, which can be accessed by clicking on Help on the main menu,

or in most places via a right click or by pressing the F1 key (context sensitive).

Design-Expert 7.1 User’s Guide General One-Factor Tutorial – Part 1? 1

2 ? General One-Factor Tutorial – Part 1 Design-Expert 7.1 User’s Guide

Design the Experiment

We will assume that you are familiar with the graphical user interface on your computer and the use of a mouse. Start the program by finding and double clicking on the icon for Design-Expert. You will then see the main menu and icon bar.

Click on File in the main menu. (Unavailable items are shown in a secondary color.) (If you prefer using the keyboard, press the Alt key and underlined letter, in this case Alt

Design Expert说明书

Design Expert说明书

F, simultaneously.)

File menu

Select the New Design item with your mouse. (The blank-sheet icon on the left of the toolbar is a quicker route to this screen. If you’d like to check this out, press Cancel to re-activate the tool bar.) You should now see four tabs on the left of your screen. The Factorial tab comes up by default. Select General Factorial for this design because the factor is categorical. (If your factor is numerical, such as temperature, then you would use the One Factor option under the Response Surface tab.) Leave the number of factors at its default level of 1 and then click on Continue

Design Expert说明书

Design Expert说明书

.

General Factorial design

Enter the Design Parameters

Type Bowler as the name of the factor. Tab to the Units field and enter Person . Then Tab to the Levels field and enter 3. Click on the Treatments field and enter Pat , Mark , and Shari .

DX71-02A-Gen1Factor-P1 Rev. 4/15/07

Design Expert说明书

Design Expert说明书

Design Expert说明书

Design Expert说明书

Design Expert说明书

Design Expert说明书

Design-Expert 7.1 User’s Guide General One-Factor Tutorial – Part 1?

3

General Factorial design-builder dialog box – completed

Leave Categoric contrasts at its default setting of Nominal which will allow names to be entered. The optional contrast for “ordinal” applies to treatments which can be placed in numerical order. This obviously is not what’s needed for this case, but if for example, our treatments were defined by something quantitative, such as height of

bowler, you’d chose this alternative contrast to generate statistical analysis that would be more consistent with response surface methods (RSM). Reminder: As we’ve already suggested, if you are setting up an experiment on a numeric factor, consider doing so via the One Factor design on Design-Expert’s Response Surface tab.

Press Continue to specify the remaining design options. In the Replicates field, which becomes active by default, type 6 (each bowler bowls six games). Tab to the Blocks field but leave it blank. Design-Expert now recalculates the number of runs for

this experiment: 18.

Design options entered

Press Continue . Let’s do the easy things first. Leave the number of Responses at the default of 1. Then click on the Name box and enter Score . Tab to the Units field and enter Pins

.

Response name dialog box – completed

4 ? General One-Factor Tutorial – Part 1 Design-Expert 7.1 User’s Guide

At this stage one can skip the remainder of the fields and continue on. However, it will be good to gain an assessment of the power of your planned design of experiment. In this case the bowling captain does not care if averages differ by less that 10 pins and his records provide the standard deviation of 5. Enter these values as shown below so

Design Expert说明书

Design Expert说明书

Design-Expert can compute the signal to noise ratio – for this design: 2.

Optional power wizard – necessary inputs entered

Press Continue to see the happy outcome – power that exceeds 80 percent probability

Design Expert说明书

Design Expert说明书

of seeing the desired difference.

Results of power calculation

Click on Continue for Design-Expert to create the design and take you to the design layout window.

Save the Design

When you complete the design setup, save it to a file by selecting File , Save As . Type in the name of your choice (such as Bowling ) for your data file, which will be saved as a *.dx7 type.

DX71-02A-Gen1Factor-P1 Rev. 4/15/07

Design Expert说明书

Design Expert说明书

Design Expert说明书

Design-Expert 7.1 User’s Guide General One-Factor Tutorial – Part 1?

5

Save As dialog box

Then click on Save . Now you’re protected in case of a system crash.

Create a Data Entry Form

Go to the View menu and select Run Sheet from the menu to get a recipe sheet of your experiments. It provides space to record the responses on a printed sheet of paper. (Note: this view of the data does not allow response entry. To type results into the

program you must switch back to the home base – the Design Layout view.)

Run Sheet view (your run order may differ )

It’s not necessary for this tutorial, but if you have a printer connected, you can select File, Print and OK to make a hard copy. (You can do the same from the basic design layout if you like that format better.)

Enter the Response Data

When you do your own experiments, you will need to go out and collect the data. Simulate this by doing a File , Exit . Click on Yes if you are prompted to Save . Then re-start Design-Expert and use File , Open Design (or file open icon on the toolbar) to open your data file (Bowling.dx7). You should now see the data tabulated in the standard design layout. (If not, go to View and select Design Layout.) For an actual

6 ? General One-Factor Tutorial – Part 1 Design-Expert 7.1 User’s Guide

experiment, the runs would have been performed in randomized run order, which is the order Design-Expert defaults to. This run order will be different each time a design is created from scratch. For this example, you must enter the data in the proper order to match up with the correct bowlers, so right click at the top of the Std column and choose Sort by Standard Order

Design Expert说明书

Design Expert说明书

.

Sort runs by standard (std) order

Design Expert说明书

Design Expert说明书

Then enter the responses from the table on page one, or use the following screen.

Design Layout in standard order with response data entered

Your design layout window should now look like that shown above, except for run order. When you do your own experiments, be sure to do the runs and enter the

response(s) in randomized order. Standard order should only be used as a convenience for entry of pre-existing designs.

Save your data by selecting File , Save from the menu (or the save icon on the

toolbar). OK, now you’re backed up in case you mess up the data. That’s good, because

Design Expert说明书

enlightenment on the screen currently displayed. Be sure to play the video clips that are

DX71-02A-Gen1Factor-P1 Rev. 4/15/07

Design Expert说明书

Design Expert说明书

Design Expert说明书

Design Expert说明书

Design-Expert 7.1 User’s Guide General One-Factor Tutorial – Part 1? 7

offered for most of the detailed features – these movies make it far clearer than static,

written tutorials can on how to make use of advanced tools.

Tips on features for Design Layout screen

Try some of the features highlighted in the Design Layout tips. For example, right-click the Select

button. This allows you to control what Design-Expert displays.

Select button for choosing what you wish to display in the design layout

Press the X button at the upper-right corner of this help screen to close it down. Now, to get a feel for the bowling results, order them from low-to-high by right-clicking on the Response column and selecting Sort by This Response .

8 ? General One-Factor Tutorial – Part 1

Design Expert说明书

Design Expert说明书

Design-Expert 7.1 User’s Guide

Sorting on a response column (also works for factors)

This is a very useful feature. It works on factors as well as responses. In this case you get a quick ‘heads-up’ that the highest games were bowled by Mark.

Analyze the Results

Next, begin the data analysis. Under the Analysis branch of the program (on the left side of your screen), click on the Score node. The Transform dialog box will be displayed in the main window of Design-Expert on a progressive tool bar. You will click these buttons from left to right and do the complete analysis. It’s a very easy process. The first dialog box gives you the option of selecting a transformation for the

Design Expert说明书

Design Expert说明书

response, which may improve the statistical properties of the analysis.

Transformation button – the starting point for the statistical analysis

DX71-02A-Gen1Factor-P1 Rev. 4/15/07

Design Expert说明书

Design Expert说明书

Design Expert说明书

Design Expert说明书

Design-Expert 7.1 User’s Guide General One-Factor Tutorial – Part 1? 9

If you need some background on transformations, first try the Tips. Then, for all the details, go to the Help command on the main menu and select Contents. Search on “transformations.” The program provides some data-sensitive advice at the bottom of the screen. In this case, it does not indicate the need for a transformation, so press ahead with the default of None by clicking on the Effects button.

Examine the Analysis

By necessity, things get a bit statistical from here on out. If this becomes intimidating, we advise you attend a basic class on regression, or better yet, a workshop on DOE such

as Stat-Ease’s Experiment Design Made Easy.

Effects button results

The really important outputs on the effects are the F-value and associated probability (“Prob>F”). In this case, there’s a very small probability, near 0.06%, that the

differences in bowling averages (the term “A-Bowler” designated “M” for model) are due to chance variation (the term “Pure Error” labeled “e” for error, generated by the within-bowler multiple games). In other words, it appears at this stage that the difference between bowlers is significant.

To get more details press the ANOVA (Analysis of Variance) button. Notice that

Design-Expert verifies that the results are significant.

ANOVA results (annotated), with context-sensitive Help enabled via right-click menu

10 ? General One-Factor Tutorial – Part 1 Design-Expert 7.1 User’s Guide

Now select View , Annotated ANOVA . Notice that all the comments disappear so you can make a clean printout for statistically-savvy clients. Re-select View , Annotated ANOVA to ‘toggle’ back all the helpful hints. Before moving on, take the first hint: “Use your mouse to right click on individual cells for definitions.” For example, try this on the p-value of 0.0006 as shown above (select Help off the pop-up menu). There’s a wealth of information to be brought up from within the program with a few simple keystrokes: Take advantage!

Design Expert说明书

Design Expert说明书

Next you see a section of the output that reports various summary statistics.

Summary statistics

The annotations tell you what you need to know, but don’t be shy about clicking on a number and getting on-line Help via a right-click (or try the F1 key). In most cases you will then get helpful advice on the particular statistic.

Now click on the scroll down arrow (at the bottom right side of screen) until you get to

Design Expert说明书

Design Expert说明书

the next section of output illustrated below.

Coefficient estimates

Unless you’re a statistician, this detailing of model terms and their confidence intervals (“CI”) probably will be of little value, although you may notice the intercept is simply the mean score of the bowlers. You may wonder why only two terms, A1 and A2, are provided for a predictive model on three bowlers. It turns out that the last model term, A3, is superfluous because it can be inferred once you know the mean plus the averages of the other two bowlers.

Now continue on to the next section labeled “Treatment Means.”

DX71-02A-Gen1Factor-P1 Rev. 4/15/07

Design Expert说明书

Design Expert说明书

Design Expert说明书

Design Expert说明书

Design Expert说明书

Design-Expert 7.1 User’s Guide General One-Factor Tutorial – Part 1?

11

Treatment means

Here you see the averages for each of the three bowlers. As you can see below, these are

compared pair-wise in the following part of the ANOVA report.

Treatment means

You can conclude from the treatment comparisons that

? Pat differs significantly (worse!) from Mark (1 vs 2)

? The 2.5 pins mean difference between Pat and Shari (1 vs 3) is not significant

(nor is it considered important by the bowling team’s captain – recall in the design specification for power that a 10-pin difference was the minimum of interest) ? Mark differs significantly (better!) from Shari (2 vs 3).

Before pressing ahead, notice the floating Bookmark too and see it work by pressing Top

.

Handy bookmarks for long reports

12 ? General One-Factor Tutorial – Part 1

Design-Expert 7.1 User’s Guide

Analyze Residuals

Click on the Diagnostics button to bring up the normal plot of residuals. Ideally this will be a straight line, indicating no abnormalities. If you’ve got a pencil (or whatever), hold it up to the graph. Does it cover all the points? Yes – so then it passes the “pencil test” for normality. You may reposition the line by dragging the line (place the mouse pointer on the line, hold down the left button, and move the mouse) or its “pivot point” (the round circle in the middle). However we do not recommend you bother doing this – the program generally puts the line in the ideal place.

Notice that the points are coded by color to the level of response they represent – going from cool blue for lowest values to hot red for the highest. For example, the red point is Mark’s outstanding 195 game. Pat and Shari think it ought to be thrown out because it’s too high. Is this fair? Click this point so it will be highlighted on this and all the other

Design Expert说明书

Design Expert说明书

residual graphs available via the Diagnostics Tool (the ‘floating’ palette on your screen).

Normal probability plot of studentized residuals (195 game highlighted)

Design Expert说明书

Notice on the Diagnostics Tool that “Studentized” is checked on by default. This converts raw residuals, reported in original units – ‘pins’ of bowling for this case, to a dimensionless number based on standard deviations, which comes out in plus or minus scale. More details on studentization can be found Help. The raw residuals can be displayed by un-checking this default mode on the Diagnostics Tool. Check it out! However, when some runs have greater leverage (another statistical term to look up in

Help!), only the Studentized form of residuals will produce valid diagnostic graphs. For example, if Pat and Shari succeeded in getting Mark’s high game thrown out (don’t worry – they won’t!) then each of his remaining five games would exhibit a

DX71-02A-Gen1Factor-P1 Rev. 4/15/07

Design Expert说明书

Design Expert说明书

Design-Expert 7.1 User’s Guide General One-Factor Tutorial – Part 1? 13

leverage of 0.2 (1/5) versus 0.167 (1/6) for each of the others’ six games. Due to

potential imbalance of this sort, we advise that you always leave the Studentized feature checked (as done by default).

On the Diagnostics Tool , select e i vs. Pred. to generate a plot of residuals for each individual game (“e i ”) versus what is predicted by the response model. [Sidebar: Supposedly “residuals” were originally termed “error” by statisticians, but the

management people got upset at so many mistakes being made!] Let’s make it easier to see which residual goes with which bowler by pressing the down-list arrow for the Color by option and selecting A:Bowler

.

Residuals versus predicted values, colored by bowler

The size of the studentized residual should be independent of its predicted value. In other words the spread of the studentized residuals should be approximately the same for each bowler. In this case the plot looks OK. Do not be alarmed that Mark’s games stand out as a whole: The spread from bottom-to-top is not out of line with his competitors, despite their protestations about the highest score (still highlighted). Bring up the next graph on the Diagnostics Tool list – e i vs Run (residuals versus run number. (Note: your graph may differ due to randomization.)

14 ? General One-Factor Tutorial – Part 1

Design Expert说明书

Design Expert说明书

Design Expert说明书

Design-Expert 7.1 User’s Guide

Residuals versus run chart (Note: your graph may differ due to randomization) Here you might see trends due to changing conditions in the alley, tiring of the bowlers or other time-related lurking variables. (Note the pattern on your graph may differ from what we show here due to the randomized run order, but that will not be relevant to our discussions.) In this case things look relatively normal. However, even if you observed a pronounced upward, downward or shift change, it would probably not bias the outcome because the runs were completely randomized. To insure against your experiment being sabotaged by uncontrolled variables, always randomize! On the Diagnostics Tool click the Influence side. Then select Ext. Student e i

Design Expert说明书

(externally studentized residuals) to see if any points stand out.

Design Expert说明书

Design Expert说明书

Externally studentized residuals graph (your graph may differ due to random runs)

DX71-02A-Gen1Factor-P1 Rev. 4/15/07

Design Expert说明书

Design Expert说明书

Design-Expert 7.1 User’s Guide General One-Factor Tutorial – Part 1? 15

On this graph, also known as “outlier t,” we are looking for points outside the plus and minus control limits. Again, for details on the meaning of external studentization, refer to Help, but in a nutshell this graph differs from the run chart in the previous figure by excluding each run, such as the one in question by Mark, prior to calculating the residual on a scale of standard deviation. In this case, all points fall within the limits (calculated at the 95 percent confidence level). In other words, Mark’s high game does not exhibit anything more than common-cause variability, so it should not be disqualified! Since there’s no indication of abnormality, it’s OK to move on to the model graph. This will tell the story about the effect of changing bowlers. It will make a good final report to the Stat-Ease team on who they should invite as a new member. Unfortunately only one person can be chosen. /

View the Means and Data Plot

Select the Model Graphs button from the progressive tool bar to display a plot

containing all of the response data and the average value at each level of the treatment (factor). This plot gives an excellent overview of the data and the effect of the factor

levels on the mean and spread of the response.

One factor effects graph with Mark’s predicted score (mean) highlighted

The squares on this effects plot represent the predicted responses for each factor level (bowler). Click on the one representing Mark. Notice that Design-Expert displays the prediction for this treatment level (reverting back to the jargon of DOE) on the legend at the left of the graph. The vertical bars represent the 95% least significant difference (LSD) intervals for each treatment. Mark’s LSD bars do not overlap with Pat’s or Shari’s, so we can say with at least 95% confidence that Mark’s mean is significantly higher than the means of the other two bowlers.

Oh, by the way, maybe you noticed that the numerical value for the length of the LSD

bar appeared when you clicked the square for Mark. You can also click on any round

points to get the actual score. Check it out!

Pat and Shari’s LSD bars overlap, so we cannot say which of them bowls best. It seems

that they must spend a year in a minor bowling league and see if a year’s worth of games

reveals a significant difference in ability. Meanwhile Mark will be trying to live up to

the high average he exhibited in the tryouts and thus justify being chosen for the top

Stat-Ease bowling team.

That’s it for now. Save the results by going to File, Save. You can now Exit

Design-Expert if you like, or keep it open and go on to the next tutorial – part two for

general one-factor design and analysis, which delves into advanced features via further

adventures in bowling.

16 ? General One-Factor Tutorial – Part 1 Design-Expert 7.1 User’s Guide

DX71-02B-Gen1Factor-P2 Rev. 4/15/07

Design Expert说明书

Design Expert说明书

Design-Expert 7.1 User’s Guide

General One Factor Tutorial – Part 2 ? 1

General One-Factor Tutorial (Part 2 – Advanced Features)

Digging Deeper Into Diagnostics

If you still have the bowling data active in Design-Expert ? software from Part 1 of this tutorial, continue on. If you exited the program, re-start it and use File , Open Design to open your data file (Bowling.dx7). Otherwise, go back and set it up as instructed in General One-Factor Tutorial (Part 1 – The Basics). Then under the Analysis branch (you may already be here from before) click on the Score node and press the Diagnostics button.

We are going to look at a new graph on the Diagnostics Tool . Click the Influence option on this floating palette. Then bring up the DFFITS . This statistic, which stands for difference in fits, measures the change in each predicted value that occurs when that response is deleted. The larger the absolute value of DFFITS the more it influences the fitted model. (For more details on this statistic and the related deletion diagnostic, DFBETAS, see the program Help or refer to page 284 of Raymond Myers’ Classical and Modern Regression with Applications

(Boston: Duxbury Press, 1986).)

DFFITS graph – highest point clicked (your graph may differ due to random runs) Notice that one point stands out above the rest. (The pattern on your graph may differ from what we show here due to the randomized run order, but that will not be relevant to our discussions.) This is Mark’s high game, which earlier created some controversy, particularly among his competitors Pat and Shari. It falls far below a relatively conservative benchmark of plus-or-minus two for the DFFITS. Thus, taking all the