Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Ever since the proliferation of both lean and Six Sigma methodologies, the
Minitab software is often used as the primary tool for statistical analysis by
the quality professional. That does not mean that Minitab is the only software
that can do the required analysis. Programs such as SAS, SPSS, Design Ease,
Excel, and many others are just as good; however, some are more difficult to
use than the others.
In this appendix, some of the important aspects that a practitioner using
Minitab should know are discussed. The focus of the discussion is on the
worksheet and the basic statistics screens. The appendix is presented in an
outlined format because of the technical content and the art that is required.
A worksheet has been chosen to elaborate because it is a very powerful
screen and unless one understands its capability, one cannot take advantage
of the software in its entirety. Basic statistics because of the ease and the
convenience it presents. If you understand the flow of these, the rest of the
guiding screens are not difficult.
In addition, we have included a short demonstration of the MSA discus-
sion using ANOVA and the 6 panels.
Introduction to Minitab
What is Minitab?
181
2012 by Taylor & Francis Group, LLC
182 Appendix A: Minitab Computer Usage
Minitab overview
Opening worksheets
Worksheet folder
Worksheet description
Column description
Data structure and manipulation
Downloaded by [Universiti Teknologi Malaysia - Skudai] at 10:51 02 October 2013
Subset/split worksheets
Information folder
Minitab Worksheet
Opening Minitab
Menu bar
Tool bar
The gray first row is for column names only, not for data
Two columns cannot have the same name in any one worksheet
Minitab Files
There are two basic types of Minitab files
Save your work as a Minitab Project File to save all your data, graphs,
and preferences together. VERY IMPORTANT: If you save your work as a
worksheet file, you only save the data.
Minitab Windows
Session window
View statistical output and Project manager
enter session commands
(shortcut keys - ctrl + M) Manage worksheets, output
graphs, and related
documents
Data window (shortcut keys - ctrl + I)
View and edit worksheets
(shortcut keys - ctrl + D)
Toolbar view
Detached by clicking on
and dragging
Folder view
Shows all elements
Downloaded by [Universiti Teknologi Malaysia - Skudai] at 10:51 02 October 2013
of a project
Left pane shows
the folders
Right pane shows
the content of the
active folder
This function is set to change all the data in column C5 (Days) to Numeric
data, and store it back in C5.
NB: You can delete the text data before using this function, but it is not
necessary.
MinitabOverview
The following example introduces the basic functions of Minitab and
demonstrates the key navigation tools. The example includes three
worksheets:
Normally, Minitab files are saved as Minitab Projects, which have the file
extension MPJ.
Find and open the data file Unavailability.mpj by double clicking on it from
within Windows Explorer.
MinitabShow Worksheets Folder
Click on the
Downloaded by [Universiti Teknologi Malaysia - Skudai] at 10:51 02 October 2013
project manager
toolbar
To make worksheets
active double-click
the folder icon next
to the worksheet
name.
Red triangle:
denotes an
available column
description.
Hover the cursor
over the triangle
to view or double
click to edit.
Data Structure and Manipulation (1). Minitab operates using columns of data.
So, where Jan Feb Mar Apr
Location 1 289 295 300 301
traditionally we Location 2 70 73 75 76
might structure data Location 3 168 174 180 189
like this:
in columns like
these:
Using the
file again:
number 1 2 3 4 5 etc or
enter the range 1:24
NB: The colon is essential
Click OK
The info window can be useful for identifying missing data, particularly
on large worksheets.
Basic statistics
contains the most
frequently used
commands.
For correlations
For 3D graphs
Exercise: Create Individual Value plots for Old data (C1) using the two
methods shown on the last two slides.
Exercise: Take a closer look at the graphics in the top left-hand corner of the
graph windows youve created. Now go back to the data in the worksheet and
change one of the values in the Old Column. What happens to the graphics?
Accessing the Stat Guide
MINITABs StatGuide helps to interpret the results of the analyses. The
StatGuide contains two windows:
Downloaded by [Universiti Teknologi Malaysia - Skudai] at 10:51 02 October 2013
Click statistics
Click OK
Trimmed mean: The top and bottom 5% of data is ignored in
calculating this version of the mean (average).
Trimmed
mean
First and
third quartile
Double click on
old and new
Click OK
Downloaded by [Universiti Teknologi Malaysia - Skudai] at 10:51 02 October 2013
you require.
In this case, select simple
Click simple
Click OK
The scatter plot will graph point 1 of the Old data against point 1 of the New
data, then point 2 of the Old against point 2 of the New and so on.
Scatter Plots (3)
If the two processes are producing the same outputs, we would expect a
scatter plot to show a 45-degree line through the origin. The output should
look like this:
Analysis: The alignment isnt perfect, but it looks like the Old and New
processes produce about the same outputexcept for one small detail: The
scales are not the same!
To see a true representation, the X axis and the Y axis must be scaled the
same.
Click on editor
Click OK
MINITAB GraphsBrushing
Some MINITAB graphs can also be interrogated using the Brush function,
found under the menu: Editor > Brush
8
7
6
5
Process Y
4
3
2
1
0
0 1 2 3 4 5 6 7 8 9
Process X
The brush function allows you to brush over specific points on a graph,
and a smaller window then appears to give you the row numbers of those
data points.
Note that the Brush function can only be used on graphs that are up to
date.
Unavailability.MPJ
Under bars represent choose
A function of a variable
Under one Y choose cluster
Click ok
MINITAB can cluster the data by regions within the shifts. In MINITAB
press the shortcut keys CTRL-E to bring back your last dialog box. Now
enter the Categorical variables in the order Shift, then Call Centre.
Click ok
4
3
2
1
0
Shift 1 2 3 1 2 3 1 2 3
Call centre Central Eastern Western
lines
6
5 5
4 4
3 3
2
1
0
Shift 1 2 3 1 2 3 1 2 3
Call centre Central Eastern Western
Click multiple
graphs
Check in separate
panels of the same
graph
Check same scales
for graphs
Click OK
We can now superimpose the New process onto the Old process to confirm
this difference between New and Old.
Click OK
Histograms (1)
Continuing with graphical analysis: The two processes in the previous
graphs appear to be different, and so it may be worth investigating the
distributions within each process to understand their differences.
Graph > Histogram > Simple
Double click on old to
place it in graph 1
Double click on new
to place it in graph 2
Click multiple graphs
Click OK
Histograms (2)
The Multiple Graphs command enables you to plot two graphs side by side,
using the same scale for comparison.
Click in separate
panels of the same
graph
Click OK
Histograms (3)
4
It might be useful to have
Downloaded by [Universiti Teknologi Malaysia - Skudai] at 10:51 02 October 2013
Histograms (4)
0
New
8
6
4
2
0
12.86 12.88 12.90 12.92 12.94 12.96
Histograms (5)
We now get two separate histograms; one for Existing and one for New. To
view them side by side we need to follow the command we used earlier:
Editor > Layout Tool
4.5 6
Frequency
Frequency
3.0 4
1.5 2
0.0 0
128.6 128.8 129.0 129.2 129.4 128.8 129.0 129.2 129.4 129.6
Downloaded by [Universiti Teknologi Malaysia - Skudai] at 10:51 02 October 2013
Existing New
The average output for the new process looks slightly higher than
the average for the existing process.
Both distributions appear fairly normalbut only a statistical
Normality test as performed later can confirm this.
Histograms (6)
We can also get MINITAB to fit a distribution line on to the histogram.
Graph > Histogram > With Fit
0
12.86 12.88 12.90 12.92 12.94 12.96
Order Number
Day of the Week
Operator
Click OK
12.850
Old New
Click OK
12.925 12.94
12.92
New
Old
12.900
12.90
12.875
12.88
12.850
09:00 am 12:00 pm 03:00 pm 06:00 pm 09:00 pm 09:00 am 12:00 pm 03:00 pm 06:00 pm 09:00 pm
Time Time
From these box plots, immediate points for investigation stand out:
Click OK
12.950
12.925
Data
Downloaded by [Universiti Teknologi Malaysia - Skudai] at 10:51 02 October 2013
12.900
12.875
12.850
Time
am
pm
ld 0 pm
09 m
pm
am
pm
pm
pm
pm
p
0
0
0
0
00
0
:0
:0
:0
:0
:0
:0
:0
:0
:0
6:
09
09
12
03
12
03
06
09
ew
O
Old
New
12.86 12.88 12.90 12.92 12.94 12.96
Data
Make the
Access the editing select a graph graph appear
Put the active
dialogue box for item to add larger or
graph in brush
the selected item smaller using
mode
zoom option
Replicate 1 Replicate 2
10 process outputs (Randomized order)
3 operators
2 replicates
Note:
c10, c11, c12 are the columns in
which the respective data are
found IN OUR EXAMPLE. You
must have ALL data STACKED in
these columns
Enter titles
OR
Enter gage
info and
options
(Note: Industry standard set at 5.15 standard deviations, if supplier uses 6.0,
comparisons will show larger error than with industry.)
Session Window
Two-Way ANOVA Table with Interaction
Graphs
Source DF SS MS F P Gage name:
Date of study:
200 1.1
% Contribution 1.0
Repeatability 30 0. 03875 0. 001292 % Study var 0.9
% Tolerance 0.8
Total 59 2. 24912 100 0.7
Percent
0.6
Gage R&R % Contribution 0.5
0 0.4
Source VarComp (of VarComp) Part 1 2 3 4 5 6 7 8 9 10
R chart by operator By operator
Total gage R&R 0. 004437 10 . 67 0.15 1 2 3 1.1
UCL = 0.1252 1.0
Repeatability 0. 001292 3 . 10 0.10 0.9
0.8
Appendix A: Minitab Computer Usage
0.7
Reproducibility 0. 003146 7 . 56 0.05 R = 0.03833 0.6
Sample range
0.5
Operator 0. 000912 2 . 19 0.00 LCL = 0 0.4
Operator 1 2 3
Operator*part 0. 002234 5 . 37 Xbar chart by operator Operator*part interaction
1.1 1.1 Operator
1 2 3
Part-to-part 0. 037164 89 . 33 1.0 1.0 1
0.9 UCL = 0.8796 0.9 2
Total variation 0. 041602 100 . 00 0.8 Mean = 0.8075 0.8 3
0.7 LCL = 0.7354 0.7
0.6
Average
Sample mean
0.4 0.5
Source (SD) 0.3 0.4
(5 . 15*SD) (%SV) (SV/Toler) Part 1 2 3 4 5 6 7 8 9 10
Total gage R&R 0 . 066615 0 . 34306 32 . 66 68 . 61
Repeatability 0 . 035940 0 . 18509 17 . 62 37 . 02
Reproducibility 0 . 056088 0 . 28885 27 . 50 57 . 77 What does all this mean?
Operator 0 . 030200 0 . 15553 14 . 81 31 . 11
Operator*part 0 . 047263 0 . 24340 23 . 17 48 . 68
Part-to-part 0 . 192781 0 . 99282 94 . 52 198 . 56
Total variation 0 . 203965 0 . 05042 100 . 00 210 . 08
Number of distinct categories = 4
215
Downloaded by [Universiti Teknologi Malaysia - Skudai] at 10:51 02 October 2013
216
3.0SL = 3.135
Average
3. 6.
Sample mean
3.05 3.05
Should be in-control Parts 1 2 3 4 5 6 7 8 9 10
R chart by operators Keith is different? By operators
0.08 Jennifer Keith Nate 3.25
0.06 3.0SL = 0.05921
2. 0.04 3.15 5.
0.02 R = 0.02300
Sample range
0.00 3.0SL = 0.00E+00 3.05
Operators Jennifer Keith Nate
Components of variation By parts
100 % Contribution 3.25
% Study var
% Toler
1. 50 4.
Percent
Parts representative ?
of process variation 3.05
0
Gage R&R Repeat Reprod Part-to-Part Parts 1 2 3 4 5 6 7 8 9 10
Appendix A: Minitab Computer Usage
Appendix A: Minitab Computer Usage 217
Total Gage R&R < 30% (% Study Variation), and Part-to-Part to be the biggest
contributor (similar as shown).
R-Chart must be In-Control. An Out-of-Control Range Chart indicates Poor
Repeatability. Also want to see five or more levels (look across the
points) of range within the control limits.
Downloaded by [Universiti Teknologi Malaysia - Skudai] at 10:51 02 October 2013
218
MSA
Health
Gage name:
side MSA
Date of study:
Reported by: Troubleshoot
Gage R&R (ANOVA) for response side
Percent
0.6
0.5
0 0.4
Gage R&R Repeat ReprodPart-to-part Part 1 2 3 4 5 6 7 8 9 10 If only 1 operator,
R chart by operator By operator you wont get
0.15 1 2 3 1.1 these graphs
Graphical Output6 Graphs In All
Sample range
0.5
0.00 LCL = 0 0.4
Operator 1 2 3
Xbar chart by operator Operator*part interaction
1.1 1.1 Operator
1.0 1 2 3 1
1.0
0.9 UCL = 0.8796 0.9 2
0.8 Mean = 0.8075 0.8 3
0.7 LCL = 0.7354 0.7
0.6
Average
0.5 0.6
Sample mean
0.4 0.5
0.3 0.4
Part 1 2 3 4 5 6 7 8 9 10
If nested study,
you wont get this
graph
Appendix A: Minitab Computer Usage
Downloaded by [Universiti Teknologi Malaysia - Skudai] at 10:51 02 October 2013
Gage name:
Date of study:
Reported by:
Gage R&R (Nested) for response Tolerance:
Misc:
Chart Output
Destructive Test
centages (%).
100 18
% Contribution
% Study var 17
16
Percent
14
0 13
Gage R&R Repeat Reprod Part-to-part Part 6 7 8 9 10 11 12 13 14 15 1 2 3 4 5
operator Billie Nathan Steve
Appendix A: Minitab Computer Usage
Sample range
Sample mean
13 LCL = 12.68
50 Needs help
0
Gage R&R Repeat Reprod Part-to-part
Components of variation
100 % Contribution
Much better
% Study var
Percent
50
0
Gage R&R Repeat Reprod Part-to-part
Answers: Where is the variation?
R chart by operator
Range chart (each
150 1 2 3 point is the range of
UCL = 124.0 repeated
measurements of the
Sample range
100
same part): Exposes
gage repeatability,
50
R = 37.97 resoultion & control
0 LCL = 0
Xbar chart: Test
of sensitivity,
reproducibility & Xbar chart by operator
population variety 1200 1 2 3
1100 Inside control
1000
limits: Blind spot
Sample Mean
900
800 of the gage
UCL = 797.1
1700 Mean = 725.7
600
500 LCL = 654.3
400
300
0.003
0.002
UCL = 0.001416
0.001
R = 4.33E-04 Beware if 25% or more of
0.000
LCL = 0 the ranges are zero.
Downloaded by [Universiti Teknologi Malaysia - Skudai] at 10:51 02 October 2013
R chart by operator
0.15 Plateaus
1 2 3
UCL = 0.1252
Sample range
0.10
% Study
(Industry standard
% tolerance