GE MINITAB Primer

Minitab Primer
initab Primer
roduction to Statistical
ta Analysis With Minitab
GE Company Proprietary
Version 2.0 1
Minitab Primer Introduc
tion
This primer is designed to provide one with the skills necessary to effectively
employ Minitab within the Six Sigma framework. It begins with an introduction
to selected file, data manipulation, and help functions, followed by a series of
demonstrations related to transaction and service quality. The student is
encouraged to work through each demonstration, following the lead of the
instructor. Each demonstration begins with a page highlighting the Minitab
functions applied in working-through the examples. These pages reflect the
commands that the user would see on the screen while using Minitab (the
heirarchical structure of Minitab is preserved).
It is assumed that the user is familiar with basic statistics, e.g., hypothesis
testing and regression analysis. A companion primer, entitled Statistics Primer
- Introduction to Statistics Through Graphical Analysis is available on the
World-Wide-Web (GE Corporate) for those requiring a review of fundamental
statistics.
Augie Stagliano
Pittsfield, MA
October 1996
Version 2.0 2
Minitab Primer Takeaw
ays
After Completing This Training, You Will Be Able To....

• Import Data Files
• Perform Basic Data Manipulation Techniques
• Use Functions to Perform Calculations
• Construct and Interpret Various Graph Types
• Generate and Interpret Basic Statistical Information
• Apply One and Two Sample Hypothesis Tests
• Perform Simple Linear Regression
• Apply χ2 Tests and One-Way ANOVA
An Overview of Applied Statistical

Techniques
Version 2.0
Stressing Interpretation of Analytical 3
Minitab Primer File Commands
• New Worksheet • Print Window

• Open Worksheet • Get Worksheet Information
• Merge Worksheet • Display Data
• Save Worksheet
• Restart Minitab
• Exit
Version 2.0 4
Minitab Primer Help and Manip Command
Manip Commands: Help Commands:

• Sort • Contents
• Stack • Getting Started...
• Unstack • How Do I...
• Search for Help On...
Version 2.0 5
Minitab Primer Demonstration On
Basic Statistical Analysis Graphical Analysis Regression Analysis

• STAT • GRAPH • STAT
• Basic Statistics • Plot • Regression
• Descriptive Statistics • Time Series Plot • Regression
• 1-Sample t • Histogram • Fitted Line Plot
• 2-Sample t • Boxplot
• Correlation • Character Graphs
• Dotplot
• Tables
• Tally • STAT
• Chisquare Test • SPC
• Run Chart
• CALC
• Probability Distributions
• Normal
• STAT
• ANOVA
• ONEWAY
Choice of Tool Depends Upon the

Requirements of the Analysis
Version 2.0 6
Minitab Primer Demonstration One
Example: Receivables “Days-to-Collection”

Data File: days.xls
Variable: Days
Collection terms for receivables is 60 days. Payments are
entered into a data collection system in the same time-order
as they are received. Characterize this process and determine
its long term z-value and sigma. Also, test that the average
days-to-collection is equal to 50 days (Business Target).
Version 2.0 7
Minitab Primer Measure - Analyze - Improve -
Control
Histogram... Time Series Plot...
90
20 80
Frequency
70
Days
10 60
50
0 40
45 55 65 75 85 I ndex 10 20 30 40 50 60
Days
Descriptive Statistics
Variable N N* Mean Median TrMean StDev SEMean

Days 50 0 63.80 64.00 63.75 8.45 1.19
Variable Min Max Q1 Q3

Days 45.00 87.00 58.75 68.25
Receivables Process Characterization
Version 2.0 8
Minitab Primer Measure - Analyze - Improve
- Control
Days Count
45 1
48 2
49 1
53 1 16 Items Within
54 3 Yield = ???
55 1 Spec (60 days)
58 3
59 3 Inverse Cumulative Distribution Function
60 1
61 2 Normal with mean = 0 and standard deviation = 1.00000
62 1
63 4
64 4 P( X <= x) x
65 1 ??? ??? z-value (LT)
66 2
67 6
68 2 34 Items Sigma = ???
69 1
70 2 Outside of Spec
71 2
72 1
74 2
77 1
78 1
79 1
87 1
N= 50
Receivables Process - Yield & Sigma Values
Version 2.0 9
Control
Hypothesis Test of the Mean.....
Business Target: 50 Days α = 0.05

Test for Mean > 50 Days Conf. Level = 95.0%
Results.....
T-Test of the Mean
Test of mu = 50.00 vs mu > 50.00
Variable N Mean StDev SE Mean T P-Value

Days 50 63.80 8.45 1.19 11.55 0.0000
Is the Average Days-to-Pay On Target ???

Version 2.0 10
Minitab Primer Demonstration Two

• Dotplot
• Tables
• Tally • STAT
• Run Chart
• CALC
• Normal
• STAT
• ANOVA
• ONEWAY

Version 2.0 11
- Control
Example: GE Stock Data

Data File: price.xls
Variable: Price
Description: this data set contains actual daily price data for a
time period of approximately two years. The data is ordered in
its original time sequence. Characterize the data and check
for stability over time.
Version 2.0 12
- Control
Descriptive Statistics Run Chart for price

Variable: price 110
100
Anderson-Dar ling Normalit y Test
A-Squar ed: 51.523 90
p-value: 0.000
price
Mean 63.674
80
St d Dev 19.583
Var iance 383.499 70
Skewness 1.338
Kurt osis 0.296 60
n of dat a 509.000
35 45 55 65 75 85 95 105 115
Minimum 45.500
50
1st Quar t ile 49.625
Median 56.750
3rd Quar t ile 65.812 100 300 500
Maximum 109.750
95%Conf idence Int erval f or Mu
95%Conf idence Int er val f or Mu
Observation
61.969 65.380
55 60 65 Number of r uns about median: 13.000 Number of r uns up or down: 261.000
95%Conf idence Int er val f or Sigma
Expect ed number of runs: 255.475 Expect ed number of runs: 339.000
18.449 20.866 Longest run about median: 245.000 Longest run up or down: 8.000
Appr ox p- value f or Clust er ing: 0.000 Appr ox p-value f or Trends: 0.000
95%Conf idence Int erval f or Median
95%Conf idence Int erval f or Median Appr ox p- value f or Mixt ures: 1.000 Appr ox p-value f or Oscillat ion: 1.000
55.000 57.500
Results of GE Stock Price Demonstration

Version 2.0 13
Minitab Primer Demonstration Thre

• Dotplot
• Tables
• Tally • STAT
• Run Chart
• CALC
• Normal
• STAT
• ANOVA
• ONEWAY

Version 2.0 14
Minitab Primer Measure - Analyze -
Improve - Control
Example: Comparing Two Different Business Regions

Data File: receive.xls
Variables: region1, region2 (t-test)
region1, reg1$$$ (scatter diagram, correlation,
and regression)
Evaluate the relative performance of these two business

regions using hypothesis testing. Also, prepare a scatter
diagram and regression model (calculate correlation co-
efficient) using Reg1$$$ as the response variable and
Region1 as the predictor.
Version 2.0 15
Improve - Control
Hypothesis Test Results...

Two Sample T-Test and Confidence Interval
Twosample T for region1 vs region2

N Mean StDev SE Mean
region1 100 46.10 10.1 1.01
region2 100 44.48 9.84 0.98
95% C.I. for mu region1 - mu region2: ( -1.2, 4.40)

T-Test mu region1 = mu region2 (vs not =): T= 1.14 P=0.26 DF= 197
Is There a Difference in the Average Level

of Receivables Ages Between Regions 1 & 2?
Version 2.0 16
Improve - Control
Scatter Plot... Correlation Coefficient (r)...
Correlations (Pearson)
750
Correlation of region1 and reg1$$$ = 0.930
650
550
reg1$$$
450
350
250
150
20 30 40 50 60 70 80
region1
Establish a Relationship Between Response

and Predictor Before Building the Model
Version 2.0 17
Improve - Control
Fitted Line Plot...
Regression Plot
850
750
650
reg1$$$
550
450 Y =29.0826 + 9.65584X

R-Squared =0.865
350
250
150
20 30 40 50 60 70 80
region1
Version 2.0 18
Regression Analysis Improve - Control
Regression Results...
The regression equation is
reg1$$$ = 29.1 + 9.66 region1
Predictor Coef Stdev t-ratio p

Constant 29.08 18.22 1.60 0.114
region1 9.6558 0.3861 25.01 0.000
s = 38.95 R-sq = 86.5% R-sq(adj) = 86.3%
Analysis of Variance
SOURCE DF SS MS F p
Regression 1 948965 948965 625.40 0.000
Error 98 148702 1517
Total 99 1097667
Unusual Observations
Obs. region1 reg1$$$ Fit Stdev.Fit Residual St.Resid
10 45.0 381.00 463.60 3.92 -82.60 -2.13R
31 41.0 525.00 424.97 4.36 100.03 2.58R
53 75.0 739.00 753.27 11.82 -14.27 -0.38 X
64 59.0 513.00 598.78 6.33 -85.78 -2.23R
70 47.0 404.00 482.91 3.91 -78.91 -2.04R
76 23.0 251.00 251.17 9.73 -0.17 -0.00 X
78 69.0 648.00 695.34 9.67 -47.34 -1.25 X
92 20.0 176.00 222.20 10.80 -46.20 -1.23 X
95 50.0 598.00 511.87 4.18 86.13 2.22R
98 45.0 558.00 463.60 3.92 94.40 2.44R
R denotes an obs. with a large st. resid.

X denotes an obs. whose X value gives it large influence.
Version 2.0 19
Minitab Primer Demonstration Fou

• Dotplot
• Tables
• Tally • STAT
• Run Chart
• CALC
• Normal
• STAT
• ANOVA
• ONEWAY

Version 2.0 20
Minitab Primer Demonstration Four
Example: Comparing Many Different Business Regions

Data File: aging.xls
Variables: Country1 - Country5
Evaluate the relative performance of five different

business regions using boxplots and dotplots.
Version 2.0 21
Improve - Control
Boxplot Results...
200
100
AGING
1 2 3 4 5
COUNTRY
Version 2.0 22
Improve - Control
Dotplot Results... ANOVA Results...
Character Dotplot One-Way Analysis of Variance
.
.: Analysis of Variance on AGING
::
:: Source DF SS MS F p
:: COUNTRY 4 424064 106016 246.30 0.000
::
:: Error 295 126978 430
.:::
::::
Total 299 551042
::::
..:::::.
-+---------+---------+---------+---------+---------+-----AGING (1)
Individual 95% CIs For Mean Based on Pooled StDev
.: Level N Mean StDev --------+---------+---------+--------
:: :.:
. ..:::::: . .
1 60 25.93 4.87 (*-)
. :. ::::::::::.::: 2 60 13.97 16.54 (-*)
-+---------+---------+---------+---------+---------+-----AGING(2)
. 3 60 46.00 16.20 (*-)
:: 4 60 118.25 27.76 (-*)
. :::.
. :::.:.:.: : 5 60 25.53 28.66 (*-)
. .:.::::::::: :.:. --------+---------+---------+--------
-+---------+---------+---------+---------+---------+-----AGING (3)
.. : Pooled StDev = 20.75 35 70 105
. ::: . ..:: .
. : .: ... ::::::::::.::.......
-+---------+---------+---------+---------+---------+-----AGING(4)
.
:.: .
::: . : : .:
:. :.:::: :..:.::::::.:: .
-+---------+---------+---------+---------+---------+-----AGING(5)
-40 0 40 80 120 160
Version 2.0 23
Minitab Primer Demonstration Five

• Dotplot
• Tables
• Tally • STAT
• Chisquare Test • Control Charts
• Xbar-S
• CALC • SPC
• Probability Distributions • Run Chart
• Normal
• STAT
• ANOVA
• ONEWAY

Version 2.0 24
Control
Example: Invoice Disputes

Data File: chisq.xls
Variables: Process, Invoices, and Disputes
Description: this data set contains the number of invoices

issued to customers using six different processes. Invoices is the
number issued and Disputes is the number of customer issues
pending problem resolution. Determine whether the results
of this test indicate a difference in the six processes.
Version 2.0 25
Improve - Control
Results of the Six Trials... Results of the Chi-Square Test...
Expected counts are printed below observed counts
process invoices disputes
invoices disputes Total
1 54 16 1 54 16 70
2 47 13 57.15 12.85
3 52 15 2 47 13 60
4 53 8 48.99 11.01
5 49 15 3 52 15 67
6 52 2 54.70 12.30
4 53 8 61
49.81 11.19
Hypothesis Test
5 49 15 64
Ho: (O-E)2 = 0 52.26 11.74
Ha: (O-E)2 > 0
6 52 2 54
α: 0.05 44.09 9.91
ν: (n-1) = 5
Total 307 69 376
Decision Rule: ChiSq = 0.174 + 0.775 +

0.081 + 0.359 +
If p < α, Reject Ho 0.134 + 0.595 +
0.205 + 0.911 +
0.203 + 0.902 +
1.419 + 6.313 = 12.071
df = 5, p = 0.035
Is the Result Significant at the 0.05 Alpha Level?

Version 2.0 26
Improve - Control
Example: Receivables Process Control

Data File: days.xls
Variables: Days
Description: Collection terms are 60 days. Payments are

entered into a data collection system in the same time-order
as they are received. Determine whether or not the process is
in control and capable of satisfying the terms.
Version 2.0 27
Improve - Control
Results of Analysis...
Xbar and S Chart for: Days
UCL=75.18
70
Means
MU=63.80
60
LCL=52.42
50
Subgroup 0 1 2 3 4 5 6 7 8 9 10
20
Std Deviations
UCL=16.65
10
S=7.970
0 LCL=0.000
Is the Process in Control and Capable?

Version 2.0 28
Minitab Primer Conclus
ion
• State the Goal of Your Work
• Identify the Desired Output
• Collect the “Right” Data…Don’t Use Data “Just Because It’s Available”
• Select the Tool(s) that Will Deliver the Desired Results
Avoid “Over Analysis” ... Identify

Your Needs Up-front and Focus on Results
Version 2.0 29
Minitab Primer Appendices
1) Solutions to Problems Using Excel

a) Demonstration One
b) Demonstration Two
2) Formulae for Calculating Sample Size

a) Attributes Tests
b) Variables Tests
3) Minitab Tools and the Breakthrough Strategy
Version 2.0 30
Minitab Primer Appendix 1a -
Data Set.... Histogram.... Run Chart.... Excel
Payment Days His togram Run Chart: Days-to-Collection
90
1 55 25 100%
90%
85
2 72 20 80% 80
Fre que ncy

70% 75
3 69
Payment
15 60% 70
4 66 10
50%
40%
65
avg
5 77
60
30%
55
5 20%
6 70 10%
50
7 79 0 0% 45
40 50 60 70 80 90 40
8 65 Days
13
15
17
21
25
27
31
33
37
39
43
11
19
23
29
35
41
45
47
49
Days
9 64
10 63
11 67 Descriptive Statistics....
. .
. . Days Observations....
Mean 63.8
Standard Error 1.19 • Normally Distributed Data
Median 64
Mode 67 • Data Stable Over Time
• Average ≈ 64 Days-to-Collection
Standard Deviation 8.4
Sample Variance 71.3
Kurtosis
Skewness
0.44
0.07
• About 68% of the Payments Occur
Range 42 Between 56 and 72 Days
Minimum 45
Maximum 87 • About 50% of the Payments Exceed
Sum
Count
3,190
50
64 Days
ConfidLevel(95.000%) 2.34
Results of Receivables Demonstration Using Excel

Version 2.0 31
Minitab Primer Appendix 1b -
Data Set.... Histogram.... Run Chart.... Excel
day price price
Histogram
1 104.00
2 103.63
25 120.00
3 103.00
4 103.88 20 100.00
5 104.38
Frequency
6 105.00 15 80.00
Frequency
7 105.63 10 60.00 price
8 106.00
9 105.25 5 40.00
10 106.88
0
11 107.13 20.00
45.50
55.62
65.74
80.91
91.03
96.09
50.56
60.68
70.80
75.85
85.97
101.15
106.21
12 108.38
0.00
13 108.75
Bin 0 100 200 300 400 500 600
14 108.38
15 108.25
. .
. .
. . Descriptive Statistics.... Observations....
price
Mean 63.67
• Bimodal Data - Two Different Groups
Standard Error 0.87
Median 56.75 • Data Unstable Over Time
Mode #NUM!
Standard Deviation 19.58 • Descriptive Statistics Unreliable Due
Sample Variance 383.50 to Data Distribution & Instability
Kurtosis 0.32
Skewness 1.35 • Significant Event Occurred at Time “100”
Range 64.25
Minimum 45.50 • Data is Upward Trending After Time “100”
Maximum 109.75
Sum
Count
32410.24
509
• The Data Set is GE Stock Price and the
Confidence Level (95.000%) 1.70 Significant Event is a Stock Split
Graphical Analysis Reveals Unusual

Version 2.0 32
Minitab Primer Appendix 2a - Attributes
Sample Size
Estimating Sample Size for Attributes
n - sample size
( )
2 Ζα/2 2
n= ·p·q Ζα/2 - Z-value for Desired Confidence Level
ω ω - Desired Precision Width
p - Population Proportion (Use 0.5 if Unknown)
q - Complement of p, i.e., (1- p)
Example: How large a sample size is required to estimate the proportion of

unpaid invoices with a margin of error of +/- 4% at a 95% confidence level?
( 2 · 1.96
)
2
n= · 0.5 · 0.5 = 600.25 601
0.08
Version 2.0 33
Minitab Primer Appendix 2b - Variables
Sample Size
Estimating Sample Size for Variables
n - sample size
( )
2
2 Ζα/2σ Ζα/2 - Z-value for Desired Confidence Level
n=
ω ω - Desired Precision Width
σ - Standard Deviation
Example: How large a sample size is required to estimate the average value of
unpaid invoices with a standard deviation of $3.50 within a margin of error of
+/- $1.00 at a 90% confidence level?
( 2 · 1.65 · $3.50
)
2
n= = 33.35 34
$2.00
Version 2.0 34
Minitab Primer Appendix 3 - MAIC
Tools
A Sampling of Statistical Tools to Apply With the Breakthrough Strategy...
Measure Analyze Improve Control
• Histograms • Hypothesis Tests • ANOVA • Run Charts
• Run Charts • Boxplots • Hypothesis Tests • Control Charts
• Descriptive • Dotplots • Regression • Confidence

Statistics Analysis Intervals
• Scatter Plots
• Dotplots • DOE
• Correlation
• Boxplots Analysis
• Regression
Analysis
Use Tools Creatively....But Avoid “Force

Version 2.0
Fitting” 35

GE MINITAB Primer

Uploaded by

Copyright:

Available Formats

GE MINITAB Primer

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

GE MINITAB Primer

Uploaded by

Copyright:

Available Formats

Minitab Primer

After Completing This Training, You Will Be Able To....

An Overview of Applied Statistical

• New Worksheet • Print Window

Manip Commands: Help Commands:

Basic Statistical Analysis Graphical Analysis Regression Analysis

Choice of Tool Depends Upon the

Example: Receivables “Days-to-Collection”

Variable N N* Mean Median TrMean StDev SEMean

Variable Min Max Q1 Q3

Business Target: 50 Days α = 0.05

T-Test of the Mean

Test of mu = 50.00 vs mu > 50.00

Variable N Mean StDev SE Mean T P-Value

Is the Average Days-to-Pay On Target ???

Basic Statistical Analysis Graphical Analysis Regression Analysis

Choice of Tool Depends Upon the

Example: GE Stock Data

Descriptive Statistics Run Chart for price

Results of GE Stock Price Demonstration

Basic Statistical Analysis Graphical Analysis Regression Analysis

Choice of Tool Depends Upon the

Example: Comparing Two Different Business Regions

Evaluate the relative performance of these two business

Hypothesis Test Results...

Twosample T for region1 vs region2

95% C.I. for mu region1 - mu region2: ( -1.2, 4.40)

Is There a Difference in the Average Level

Establish a Relationship Between Response

450 Y =29.0826 + 9.65584X

Predictor Coef Stdev t-ratio p

s = 38.95 R-sq = 86.5% R-sq(adj) = 86.3%

R denotes an obs. with a large st. resid.

Basic Statistical Analysis Graphical Analysis Regression Analysis

Choice of Tool Depends Upon the

Example: Comparing Many Different Business Regions

Evaluate the relative performance of five different

Basic Statistical Analysis Graphical Analysis Regression Analysis

Choice of Tool Depends Upon the

Example: Invoice Disputes

Description: this data set contains the number of invoices

Decision Rule: ChiSq = 0.174 + 0.775 +

Is the Result Significant at the 0.05 Alpha Level?

Example: Receivables Process Control

Description: Collection terms are 60 days. Payments are

Is the Process in Control and Capable?

• State the Goal of Your Work

• Identify the Desired Output

• Select the Tool(s) that Will Deliver the Desired Results

Avoid “Over Analysis” ... Identify

1) Solutions to Problems Using Excel

2) Formulae for Calculating Sample Size

3) Minitab Tools and the Breakthrough Strategy

Fre que ncy

Results of Receivables Demonstration Using Excel

Graphical Analysis Reveals Unusual

Example: How large a sample size is required to estimate the proportion of

Measure Analyze Improve Control

• Histograms • Hypothesis Tests • ANOVA • Run Charts