Measure

Phase
1 Process
Mapping

2 Operational Definition & Data

Collection

3 Basic Statistics & Graphical Analysis of

Data

4 Measurement Systems
Analysis

5 Process
Stability

Soubhagya Ranjan Behera 6 Process

Baseline

10-Dec- 1
CTQ Identification
 Critical to Quality (CTQs) help to qualify and quantify Voice of
what is important to the customer Customer
“I want this product delivered
 If you cannot measure something, how do you know if
faster”
you have
improved it?
 They help specify requirements – “must-have” or “must-be”
attributes or ultimate satisfiers Critical to
 They can be measured Quality
The time taken from when the
 Measurement system must be acceptable customer places an order to when
 They establish aspecification
 Customer target limits will enable the product is delivered to the
performance customer should be less than 3
assessment days.
 Acceptable range of performance (process CTQs are (1) important to the
capability) customer and (2)
measurable
2
Importance of Data
 It is important to develop a fact based approach to Ask
Collect
make decisions and solve business problems (rather question
s Data
than guessing at the answer)
 Decide what data we need to answer these
questions?
 Collect the data. Sometimes data may already exist but
at other time we may need to put a process to collect
the data Answer
Analyze
 Analyze the data, using graphical and statistical question
s Data
techniques
 Answer the questions based on analysis of the data
 Make decisions based on the answers. This will help Draw
you convince others on why you are recommending a Conclusion
particular course of action. s

Using data to answer

questions is the Six Sigma 3
Methods to collect data
 Manufacturing
 Electronic sensors
 Manual
o Scrap buckets – in which people count and tally the information at the end of every day
o Check sheets – in which operators check that things have been done or defects have
occurred

 Transactional
 SAP – for example, time stamps
 Manual data collection is often required – consider the use of check sheets
 MES data

If data does not exist to answer your questions, you need to

put
a data collection plan how to collect.

4
How much data to collect?
 To be a representative of the population sample selection should be:
 Impartial- biasness means treating all equal
 Objective- facts and evidence
 This means that each item of the population must have an equal
chance of getting selected.
 The method which ensures “Equal Chance” for each item of the
population
is called as……“Random Sampling”
 Selecting items only from top of each box is not random sample,
as the items at the bottom of the box have no chance of getting
selected.
 Sequential Sampling: Similar to Multiple Sampling. Useful when
Process is
“infinite”
 Samples may be selected Randomly or Sequentially.
5
Check Sheet / List
 What is it?
 A tool for collecting data in a consistent form.
 Provides an easy, structured way of recording data as it is
collected
 Assures data will be recorded in similar manner
 Formats can be designed for various needs
 Most commonly arranged in rows and columns or matrix
 Used by improvement teams to gather data
 Benefits
 Collects data in a systematic and organized manner
 To determine source of problem
 To facilitate classification of data (stratification)
6
Data Collection Plan
 A data collection plan lists the details about how the data will be collected.
 This plan can then be handed over to the team members to execute the data collection plan and provide to the
rest of
the team members.
What Operational Definition Who When Where How Much

Cycle Time from order entry to Shyam Beginning SAP /MES 10%
Time order fulfilled in days of every records random
month sample
% Number of defects divided Shiva Every shift Scrap buckets 1 hour
Defects by number of production every shift
pieces (including rework) change

7
Measure
Phase
1 Process
Mapping

2 Operational Definition & Data

Collection

3 Basic Statistics & Graphical Analysis of

Data

4 Measurement Systems
Analysis

5 Process
Stability

6 Process
Baseline
8
Types of Data
 There are two basic types of data:
 Continuous
 Attribute (or Discrete)
 Continuous data are usually measured using an instrument and any value is
possible:
 Calipers (length)
 Thermometer (temperature)
 Clock (time)
 Attribute or discrete data come in discrete units that cannot be subdivided.
 Number of customer complaints
 Good/Bad service
 The
 type of data
Which we have
statistical testsdetermines:
may be
appropriate
 How much data we need to collect We need to understand types of data to determine
which
tests to use and how much data to collect
9
Types of
Continuous Data-that take Attribute
Data value Data
 Binary: Only 2 values are possible
 Good / Bad
 Interval Data(where each point put in equal
 True / False
distance): Can compute difference between  Yes / No
two measurements but ratio does not make  Defective / Not Defective
sense  Nominal Data: More than 2 values but there is
 Temperature (degree Centigrade),
no ordering

 Nationality
 Ratio Data: Can compute difference and ratio  Religion
of data, t  Occupation
 Weight
 Ordinal Data: More than 2 values and ordered
 Age
or ranked
 Height  Customer satisfaction scores on a scale of
1-10
 Grades on a report card
 Service quality rating

10
Basic Statistics
 Descriptive statistics can be used to summarize and describe the process in numbers:
 Central location (Mean, Median, Mode)
 Variability (Range, Standard Deviation, Quartiles)
 The central location is used to determine the central value of the data. It can be used to determine for
example if the process output is close to a target.
 The variability measure can be used to quantify the amount of spread or variation in the data from the central
target
value.

Population parameters (µ and

σ)

Population Sample Statistics (𝑥 and

s)
Sample

11
 Measures for central tendency (single value that represent group of
value)
Measure Description
Mean (Average) Average of a set of values

Median
Formula: (N+1)/2th value, where N is
the number of data points even

Mode Most frequently occurring value in the data set

σ 𝑥𝑖
𝑥= Most commonly used measure for central
𝑛 tendency
12
Measures for variation
Measure Description
Range Numerical distance between the highest and lowest values in the
data set
Range: Max - Min
Standard Deviation Square root of the average squared deviation of each individual value from the mean
Square of the standard deviation is variance

𝑠= σ 𝑥 𝑖 −𝑥 2 Most commonly used measure for spread

𝑛−1

13
Normal Distribution
 The nature of the distribution of the data can provide important information
to understand the characteristics of the underlying process from which the
data is obtained.
 Many natural phenomena and man-made processes are distributed normally,
or can be represented as normally distributed
 The normal distribution has certain special properties:
 The normal distribution is characterized by two parameters (mean and
the
standard deviation)
 The mean is the central value where the distribution is located (0 in -
the figure on the right)
 The standard deviation indicates how “fat” the distribution is (1 in the
figure on the right)
 The area under the normal curve gives the probability of the data falling The normal distribution is
in also known as the
that zone (68% of the data lies between -1 and +1) Gaussian distribution or bell
 If the mean and standard deviation are 0 and 1 respectively, then curve
it is called the standard normal curve (Z).
14
Minitab

Introduction to Minitab
 Minitab is designed for statistical analysis and can be easy to
use once you know where to get help
 We will be using Minitab software throughout the learning
process
 Approach to use Minitab:
 Manipulate your data in Excel and then copy and paste
the data into Minitab.
 Use Minitab to analyze your data and then copy and
paste
the analysis to Power Point.
 Make sure to save your analysis so that you can
revisit them later.
 Main
 sections:
Menu bar (access different tools &
functionality)
 Session window (all your text output goes Minitab is the standard tool for data analysis in the
here) Six Sigma world
 Worksheet (enter your data here in columns)
15
Minitab

Introduction to Minitab
 Files:
 Minitab analysis can be saved to a project file (*.mpj) - recommended
 Only the raw data can be saved without the analysis in an worksheet file
(*.mtw)

A B D

 Menu Bar
 Reopen the last dialog box you were working on (Ctrl-E)
 Open the session window
 Open the worksheet
 Open the graphs folder

 Help Menu
 The help menu is very powerful within Minitab and we encourage you to use this to
better understand the tools and features of Minitab.

16
Minitab

Introduction to Minitab
 Copying data into Minitab – make sure to copy the header
into the gray bar
 Note the column titles (C1-T indicates that that column contains
only
text)
 Column C2 does not have –T next to it, so that column contains
only numbers.
 If you don’t copy the header information correctly like in columns
C4 and C5, the sales number column is incorrectly tagged as a
text column.
 Minitab needs to know the column type so that it can feed the
right data to the right analysis. For example, if an analysis expects
numbers only, then Minitab will not display any columns that
contain text.
 Minitab likes data to be in column format rather than row format.
So, format your data accordingly.
 To re-format data you can use one of the following options:
o Data > Stack Columns
o Data > Unstack Columns
o Data > Code
o Data > Change Data Type

17
Minitab

Basic Statistics
1. Click on Stats > Basic Statistics > Display Descriptive 2
Statistics
2. Double click on the Sales column to select it into the
Variables box (or single click on the sales column and then
click on Select button)
3. Click on Statistics to select the values you want to
display or just accept the defaults.
4. Click on the OK button to generate the analysis results.
Your analysis output is sent to the session window. 3
• analysis
5. The The number of data points (N)
shows: 4
• The number of missing data points in the column
(N*)
• The mean value
• The standard deviation value
• The median value
• The minimum, maximum values
• The first and the third quartiles

18
Minitab

Introduction to Minitab graphical analysis

19
Minitab

Graph the distribution of data

20
 Minitab

Histogram
Histogram of Diameter
 Appropriate for Diagnostic Report
Look for these patterns:
Moderate
 Continuous data Normal Right Skew

 Larger sample size (> about

50) Truncated
Normal
Severe
Right Skew

 What
Canisbethe
used to location and
central
spread of data
Moderate

 View shape of the distribution (is it Bimodal Left Skew

skewed)
 Check if the distribution is Uniform Outliers

normally distributed Examine key characteristics of your process:

• Center: Locate the peak (cluster of tallest bars), which typically represents the process

 Check if data is unimodal or center.

• Variability: Assess the spread of the data to determine how much your process varies.
• Shape: Assess the overall shape of the data and look for unexpected patterns.

multimodal •Outliers: Identify outliers (isolated bars at the ends of the graph), which can
indicate problems in your process.

 Are there gaps or outliers in the data

(an observation that lies an
abnormal distance from other
values in a random sample from
a population
21
Minitab

Steps to create Histogram

 With Assistant
 Click on Assistant > Graphical Analysis >
Histogram
 Select all data are in one column
 Enter the Y column name as Diameter
 Click on Add normal distribution curve
Histogram of Diameter

 Click on OK to generate the histogram

 Without Distribution of Data
Summary Report

Examine the center, shape, and variability.

Descriptive Statistics

Assistant
 Click on Graph > Histogram (select With
12 N 40
Mean 1.3017
StDev 0.051525
10 Minimum 1.1886
Maximum 1.3999

Fit) 8

 Double click on the Diameter column

Frequency
6

 Click on OK
4

0
1.20 1.25 1.30 1.35 1.40
Diameter

22
Minitab

Individual Value Plot

Individual Value Plot of Diameter by Stage
 Appropriate for Diagnostic Report
Look for differences in:

 Continuous data
1.40
Center

 Smaller sample size (< about 1.35

50)
Variability
 Can
 be used tomultiple groups
Compare 1.30

 Are the central locations similar across

groups 1.25
Center and Variability

 Are the spread similar across groups

1.20

 What is the confidence interval for

Shift 1
means Shift 2

 Are there outliers in the data Compare key characteristics of your process across multiple samples of data:
•Center: Compare the location of the mean symbols and their 95%
confidence intervals, which provide a range of likely values for the
mean. These values are estimates of the process center for each sample.
• Variability: Compare the spread of the data values, which indicate how much each sample varies.

If you find meaningful differences among the samples, use this information to help guide your process
improvement efforts.

23
Minitab

Steps to create Individual Value Plot

 With
Assistant
 Click on Assistant > Graphical Analysis >
Individual
Value Plot
 Select all data are in one column
 Enter the Y column name as Diameter
 Specify Number of X columns as 1 Individual Value Plot of Diameter by Stage
Summary Report

 Enter the X1 column as Stage Distribution of Data by Group

Compare the center and the variability across samples. Identify any outliers (marked in red).

 Click on OK to generate the plot

1.40

 Without 1.35

Assistant
 Click on Graph > Individual Value Plot (With 1.30

Diameter
Groups) 1.25

 Double click on the Diameter column as

1.20

Shift 1 Shift 2

Variable column Statistics

Stage

Shift 1
Shift 2

 Double click on Stage as the Categorical

N
Mean 20
StDev
Minimum 20
Maximum 1.3003

Variable column 1.3030

0.061753
0.040400
1.1886
1.2342

 Click on OK
1.3999
1.3819

24
Box Plot
 Appropriate Boxplot of Diameter by Stage
Diagnostic Report
for Continuous data Look for differences in:
1.40
 Larger sample size (> about Center

50)
 Can beSubgroups
used
1.35

to  Compare multiple groups Variability

 Understand central location and 1.30

spread
 Find shape abnormalities 1.25

 Compare shapes between Center and Variability

subgroups
 Construction: 1.20

 Find potential
The line in theoutliers
middle of the box represents the
Shift 1
median Shift 2

 The lower edge of the box represents the first quartile Compare key characteristics of your process across multiple samples of data:
•Center: Compare the mean symbols, which are estimates of the
Q1 process center for each sample. You can also compare the location of
the interquartile range boxes, which represent the middle 50% of the
 The upper edge of the box represents the third quartile data.
•Variability: Compare the lengths of the boxes and the whiskers

Q3 (lines that extend from each end of the box), which indicate how
much each sample varies.

 The whiskers extend from the edges of the box to the If you find meaningful differences among the samples, use this
information to help guide your process improvement efforts.
furthest data point within a range of 1.5x(Q3-Q1)
 The stars represent any data points beyond the range
of 1.5x(Q3-Q1) – potential outliers
25
Minitab

Steps to create Box Plot

 With
Assistant
 Click on Assistant > Graphical Analysis > Box
Plot
 Select all data are in one column
 Enter the Y column name as Diameter
 Specify Number of X columns as 1
 Enter the X1 column as Stage Boxplot of Diameter by Stage
Summary Report

 Click on OK to generate the plot

Distribution of Data by Group
Compare the center and the variability across samples. Identify any outliers.

 Without 1.40

Assistant
 Click on Graph > Box Plot (With Groups)
1.35

Diameter
1.30

 Double click on the Diameter column as Variable 1.25

column
1.20

Shift 1 Shift 2
Stage

 Double click on Stage as the Categorical Variable Statistics

N
Mean
StDev
Shift 1
20
1.3003
0.061753
Shift 2
20
1.3030
0.040400

column
Minimum 1.1886 1.2342
Maximum 1.3999 1.3819

 Click on OK

26
Graphical Summary
Graphical Summary of Diameter by Stage
Diagnostic Report
 Appropriate for Distribution of Data
Look for differences in:

 Continuous data Shift 1

Center

 Can be used to
 Compare multiple groups Variability

 Understand central location and

spread Shift 2
Center and Variability

 Compare distribution shapes

 Check if the data is normal
 One click to understand basic statistics
& graph
1.20 1.25 1.30 1.35
1.40

Compare the center and the variability across the samples. If you
find meaningful differences, use this information to help guide
your process improvement efforts.

This is one of the most used and most

useful commands in Minitab
27
Minitab

Steps to create Graphical Summary

 With Assistant
 Click on Assistant > Graphical Analysis > Graphical
Summary
 Select all data are in one column
 Enter the Y column name as Diameter
 Specify Number of X columns as 1
 Enter the X1 column as Stage
 Click on OK to generate the plot
 Without Assistant
 Click on Graph > Box Plot (With Groups)
 Double click on the Diameter column as Variable column
 Double click on Stage as the Categorical Variable column
 Click on OK

28
Minitab

Bar Chart
Bar Chart of Diameter D
Summary Report
 Appropriate for
Distribution of Data
Compare the height of the bars across
 Attribute data 20
categories.

 Binary or
ordinal 15

 Can be used to

Dat
10
 Compare relative

a
proportions
5

0
Undercut Oversize Inclusion Burr Chatter Discolored Other
Diameter D

Total: 46 Average: 6.5714 Minimum: 2 Maximum: 18

29
Minitab

Steps to create Bar Chart

 With
Assistant
 Click on Assistant > Graphical Analysis > Bar Chart
 Select Summarized values in a table
 Select Diameter Defect for the Category names
column
 Select Dia Defect Count for Summarized value
column Bar Chart of Diameter D
Summary Report

Distribution of Data

 Leave the Order bars by as Default

Compare the height of the bars across categories.

 Without
20

 Click
Click on
Assistant on OK to generate
Graph the plot
> Bar Chart
15

 Select the option Values in a table(Simple)

Data
10

 Select Dia Defect Count as the Graph variable 5

 Select Diameter Defect as the Categorical 0

Undertcu rt Ove s ze r i nc us on I l i Bu rr tt r iChal er
D sco o ed

variable
O he
Diameter D
Total: 46 Average: 6.5714 Minimum: 2 Maximum: 18

 Click on OK

30
Minitab

Pie Chart
Pie Chart of Diameter D
Summary Report

Diameter D
Undercut
4.3% Oversize
Inclusion
8.7%
Burr
Chatter
26.1% Discolored
other
6.5%

 Can be used to
 Compare relative
proportions 4.3%

10.9%

39.1%

Total count = 46

Assess the relative contribution of each category to the whole. Compare the size of the slices and
use the percentages to assess the contribution of each category.

31
Minitab

Probability Distribution Plot

 Appropriate for
 Continuous data

 Can be used to
 View or calculate probability or
inverse probability values for a given
distribution Distribution Plot

 Let’s create a probability plot for a 0.09

Normal, Mean=15, StDev=5

normal distribution with a mean value 0.08

0.07
of 15 and standard deviation of 5. 0.06

 Click on Graph > Probability 0.05

Densit
Distribution Plot
0.04

y
0.03

 Click on View Single and OK 0.02

 Select Normal distribution and specify 0.01

mean 0.00
0 5 10 15 20 25 30
X
= 15 and Stdev = 5.
 Click on OK
32
Minitab

Probability Distribution Plot

 Let’s calculate the probability of values
greater than 20 for a normal distribution
with a mean value of 15 and standard
deviation of 5.
 Steps
 Click on Graph > Probability
Distribution Plot Distribution Plot
 Click on View Probability and OK Normal, Mean=15, StDev=5

 Select Normal distribution and

0.09

0.08

specify 0.07

mean = 15 and Stdev = 5. 0.06

 Click on Shaded Area and select the 0.05

Density
define shaded area by X value 0.04

 Select the Right Tail 0.03

0.02
 Select the X value of 20. 0.01
0.1587

 Click on OK 0.00
15 20
 The shaded area (the probability of X

getting values greater than 20) is

15.87% 33
Measure
Phase
1 Process
Mapping

2 Operational Definition & Data

Collection

3 Basic Statistics & Graphical Analysis of

Data

4 Measurement Systems
Analysis

5 Process
Stability

6 Process
Baseline
10-Dec- 34
Measurement Systems Analysis (MSA)
Accuracy– how closes results are to known value
 Measurement System Analysis is a set of experiments Precision – how close results are
to determine if the measurement system is
acceptable
 Accurate (on target)
 Precise (little variability)
 There is an MSA for each of the two types of data
 Continuous (variables)
 Attribute (For creating control chart)
 The objective of this study is to determine whether
actions are
necessary to repair or replace the measurement
system
 To perform an MSA, first determine the source of data and
where measurement errors could occur. Set up MSA around
the most significant sources of error
 Most likely areas for measurement error:
 Inputting the data
 Manipulating the data
 Reporting the data
10-Dec- 35
MSA Methods

In the
Scope
of Green
Belt
In the
Scope of
Black Belt

36
Bias is the
difference between
average of
measured values to
respective true
values
Example of to calculate R R

1-Parts name
2- part number -Abc
3-Gauge name and characterstic- what you want to
check
4-nominal customer NAME
5- Range4.5 to 5
6-DATA performed by date
Range = Maximum- minimum
Total part - 10
Xp bar = AVG (Xa Xb Xc)
Rp= (Max Xp bar- Min Xpbar)
Rbar= (Ra bar +Rb bar +Rc bar)/ No of appraisers
X diff bar = Max off(Xa bar , Xb bar ,xc Bar )- Min of (Xa bar , Xb bar , Xc bar )

PART
Note ----Recent edition K value changes so only
N- no of part ,r- no of appraiser
use recent values during calcualtionk1 , k2, k3
only change those values
 Please note K1 , K2 , K3
values changes according
to d2 value , current
Ev – EQUIPMENT Variation values is according to
Tv– Total Variation below table , For Manual
Av- Appraiser variation calculation , trail ,
Pv – PART variation. appraiser and sample
normally taken like this
Steps to create attribute MSA study
 Create Design
 Click on Assistant > Measurement Systems Analysis
 Select the Setup Attribute Agreement Worksheet
 Enter the number of appraisers for this study (3)
 Enter the number of trials (2)
 Names of the appraisers (optional)
 Select Get from current worksheet (dropdown box)
 Test Items (Test column)
 Known Standard (Silver-Std)
 Value of good item (A)
RunOrder Appraisers Trials Test Items Results Standards
 Conduct Study 1 Appraiser 1 1 15 R
2 Appraiser 1 1 1 R
 The MSA worksheet is created for this study 3 Appraiser 1 1 20 A
 Ask the appraisers to measure these parts in random 4 Appraiser 1 1 11 R
5 Appraiser 1 1 16 A
order 6 Appraiser 1 1 23 A
 Enter the inspected values in the Results column 7 Appraiser 1 1 18 R
8 Appraiser 1 1 14 R
 Once all the tests are done, analyze the test results.

51
ACCEPTABLE CRITERIA OF Gauge
Analyze attribute MSA results
 The data for this analysis is given in columns Silver-
A1, Silver-A2, Silver-B1, Silver-B2. These readings
are the values obtained by the operators for
inspecting 30 parts measured 2 times.
 Click on Assistant > MSA > Analyze Attribute
Agreement Analysis
 Enter the information in the dialog box:
 Column name for the appraisers
 Column name for the test items
 Column name for the appraisal results
 Value of the item that is good (A)
 Click on OK to perform the analysis.

53
MSA (Attribute) Results
 The bar chart shows the % accuracy by appraiser.
Ideally, these values should be greater than 90%. The
overall value is 51.7% which is not acceptable.
 The misclassification rate shows the % of times the
items were incorrectly classified:
 56.8% of the time an acceptable part was
classified as rejects.
 25% of the time a rejected part was
considered acceptable.
 The figure at the bottom provides some additional
breakdown by specific items that caused the most
problems or the breakdown by appraiser (who made the
most mistakes).

10-Dec- 54
Thank You