Professional Documents
Culture Documents
ANALYSE PHASE
TCS Internal
DMAIC Roadmap
Define
Define
Measure
Measure
Identify
Project CTQs
Establish
Performance
Standard
Develop
Project Charter
Understand
As-Is
process
Prepare High
Level Process
map, SIPOC
Assess
Measurement
System
Variation
Complete
Stakeholder
analysis
Estimate
Current
Capability
Analyze
Analyze
Identify variation
using Graphical
analysis
Prioritize &
Validate
causes
Improve
Improve
Define y= f (x)
Identify
Solutions
Prioritize
And
Implement
Solutions
Measure
improvements
Control
Control
Optimize &
refine solutions
Measure
actual benefits
Close &
Hand-over project
Identify
Potential
Causes
Sampling &
Data Collection
2
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Why Analyze ?
To understand the problem and identify
root causes
To avoid actions based on intuition,
preconceived ideas & symptoms
To develop sustainable process
improvements for long term benefits
Recalibrate project scope
Establish performance goals for the
process
Find the Xs that affect Y most
3
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Input
Measures
(Xs)
Outputs
(Ys)
22 November 2013
Y=
f(x1,x2,x3,x4,,xn)
30+ variables
A
N
A
L
Y
S
E
15-20
variables
10-15
variables
5-10
variables
3-5
variables
22 November 2013
Box Plot
Validate
Causes
Hypothesis testing
Scatter Plot
Pareto Analysis
6
Copyright 2013 Tata Consultancy Services limited
22 November 2013
22 November 2013
Box Plot
Purpose:
To begin an
understanding of the
distribution of the data
To get a quick, graphical
comparison of two or
more processes
Outlier
any point outside the lower
or upper limit
Maximum Observation
that falls within the upper
limit
= Q3 + 1.5 (Q3 - Q1)
When:
First stages of data
analysis
75th Percentile
(Q3)
Median (50th
Percentile)
25th Percentile
(Q1)
Minimum Observation
that falls within the lower
limit
= Q1 - 1.5 (Q3 - Q1)
8
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Box Plot
Things to look for in a Box Plot:
Are the boxes about equal or different?
Do the groups appear normal (symmetrical box
halves and whiskers) or skewed?
Are there outliers?
Boxplots of Op1 Cycl and Op2 Cycl
(means are indicated by solid circles)
20
10
0
Op1 Cycl
Op2 Cycl
9
Copyright 2013 Tata Consultancy Services limited
22 November 2013
10
Copyright 2013 Tata Consultancy Services limited
22 November 2013
70
10
8
8
60
50
F r equency
F r equency
Data
6
5
4
40
3
2
30
20
TAT- Agent 1
TAT- Agent 2
30
40
50
60
T A T - A gent 1
70
20
30
40
50
T A T - A gent 2
60
11
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Scatter Plot
Scatter Plot tool can be used when
Both X and Y are in continuous format
If we want to associate Y with a single X
To judge the strength of relationship between Y and X
12
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Correlation
r is always between 1 & +1.
Positive value of r means direction of movement in both
variables is same
Negative value of r means direction of movement in both
variables is inverse
Zero value of r means no correlation between the two variables
Higher the absolute value of r, stronger the correlation between
Y & X
13
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Types of Correlations
r=0.9
n=30
r=0.6
n=30
r=-0.6
y-effect
n=30
x-cause
Positive Correlation
r=-0.9
n=30
r=0.0
n=30
r=0.0
y-effect
n=30
x-cause
No Correlation
No Linear Correlation
Negative Correlation
14
Copyright 2013 Tata Consultancy Services limited
22 November 2013
15
Copyright 2013 Tata Consultancy Services limited
22 November 2013
16
Copyright 2013 Tata Consultancy Services limited
22 November 2013
95
90
85
80
75
70
50
55
60
65
On-boarding Test Score
70
75
22 November 2013
18
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Pareto
What is it ?
Why use it ?
19
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Pareto - Example
Minitab Command: Stat > Quality Tools> Pareto Chart
Chart defects table: Query Type for Labels in & Total received for Frequencies
in
20
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Pareto - Example
Pareto Chart of No. of Queries rec'd
20000
Total
15000
80
60
10000
40
5000
Sub type
20
r
GE UT E ION IO N TED T ED AIL ERY PIN US the
N
O
M
T
T
U
P A
&
O
A
M
LA
LA
A
C H DIS RM ELL RE RE ION T Q RD NY
N
A
O
S
T
FO ANC A UD RDS C A ME T C A N
ES
N
I
R
Y
I
S
F
C FR CA
D
PA FA
RI
ER RD
AD
A
M
CL
CA
TO
S
CU
Total
Percent
Cum %
Percent
100
4116 3431 2709 2685 1749 800 506 269 247 234 864
23.4 19.5 15.4 15.2 9.9 4.5 2.9 1.5 1.4 1.3 4.9
23.4 42.9 58.2 73.5 83.4 88.0 90.8 92.4 93.8 95.1 100.0
21
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Box Plot
Validate
Causes
Hypothesis testing
Scatter Plot
Pareto Analysis
22
Copyright 2013 Tata Consultancy Services limited
22 November 2013
23
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Entire
Population
of Data
Sample
Analysis
Statistical
Inference
Statistics:
x, s, etc.
24
Copyright 2013 Tata Consultancy Services limited
22 November 2013
25
Copyright 2013 Tata Consultancy Services limited
22 November 2013
P < : Reject Ho
P > : Accept Ho
26
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Continuous
data
Discrete
data
Comparing
Proportions
1Proportio
n test
2Proportio
n test
Chisquare
test
Determining
statistical differences
within and between
populations
Comparing
Means
two
samples
2sample
t-test
one
sample
1sample
t-test
Comparing
Variances
multiple
samples
two
sample
ANOVA
Test of equal
variances
27
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Process Scenarios
1 Sample t-test
2 Sample t-test
ANOVA
Test of equal
variances
1-Proportion
test
2-Proportion
test
Chi-square test
To check association between variables like whether there is any association between two teams w.r.t.
their Error types.
28
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Ho
Ha
: 1 = 2
: 1 >< 2
29
Copyright 2013 Tata Consultancy Services limited
22 November 2013
30
Copyright 2013 Tata Consultancy Services limited
22 November 2013
One-sample T-test
Example:
Organization ABC is measuring the no. of days to get money from XYZ
after invoices are sent. Historical data suggests that earlier payments
were received within 25 days, however some improvement actions were
implemented. Process wanted to check whether improvement plans have
any impact on the performance.
The sample data was collected. The time taken for receiving the
payments are : 22, 23, 22, 25, 28, 27, 28, 25, 23, 21 days.
Establish whether we get money in 25 days with 95 % Confidence.
Instructions
Stat > Basic stat > 1 sample t
Enter data as: Variable C1 Days
Test Mean: 25 , Alternative Not Equal
31
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Minitab Output
T-Test of the Mean
Test of mu = 25.000 vs mu not = 25.000
Variable
Days
N
10
Mean
24.400
StDev SE Mean
2.591
0.819
-0.73
P
0.48
22 November 2013
H0: m1 = m2
Ha: m1 > < m2
33
Copyright 2013 Tata Consultancy Services limited
22 November 2013
34
Copyright 2013 Tata Consultancy Services limited
22 November 2013
35
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Alternate Hypothesis:
Ha: m1 = m2
Interpretation:
Since p < 0.05, the experienced
engineers are different from the new
engineers
Alternate Hypothesis:
Ha : m1 < m2
Interpretation:
Since p < 0.05, the experienced
engineers are taking less time than
the new engineers
36
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Hypothesis:
H0: m1 = m2 = m3 =
versus
Ha: there is at least one difference
37
Copyright 2013 Tata Consultancy Services limited
22 November 2013
ANOVA Assumption
The purpose of one-way ANOVA is to compare means. The means
of different groups of data can only be compared if the variances
within each group are statistically the same.
ANOVA has two assumptions:
Data for each group should be normal
The data sets have equal variances.
38
Copyright 2013 Tata Consultancy Services limited
22 November 2013
ANOVA: Example
A contact centre used to receive call
for different processes within
organization. The Contact Centre
head wanted to understand whether
the response time is affected by
different processes.
Response time data was collected
for the 3 processes for doing ANOVA
analysis.
Process A
Process B
Process C
3.5
6.5
3.5
5.5
4.5
5.5
5.5
7.5
6.5
4.5
6.5
5.5
5.5
4.4
6.5
5.4
6.5
39
22 November 2013
4.0
4.5
5.0
5.5
6.0
6.5
A -S quared
P -V alue
0.46
0.212
M ean
S tD ev
V ariance
S kew ness
Kurtosis
N
5.9615
0.8282
0.6859
-1.02716
1.44419
13
M inimum
1st Q uartile
M edian
3rd Q uartile
M aximum
7.0
4.0000
5.5000
6.0000
6.5000
7.0000
5.5
6.0
6.5
7.0
7.5
5.5000
9 5 % C onfidence Inter vals
0.5939
M inim um
1st Q uartile
M edian
3rd Q uartile
M axim um
5.5000
6.0000
6.0000
7.0000
7.5000
6.7240
6.0000
7.0000
6.5000
1.3671
6.3462
0.6253
0.3910
0.387879
-0.844201
13
5.9683
6.4620
Mean
M ean
S tD ev
V ariance
S kew ness
K urtosis
N
0.59
0.101
A -S quared
P -V alue
0.4484
1.0322
Mean
Median
6.0
6.2
6.4
6.6
6.8
7.0
Median
5.50
5.75
6.00
6.25
6.50
4.5
5.0
5.5
6.0
6.5
A -S quared
P -V alue
0.46
0.212
M ean
S tD ev
V ariance
S kew ness
Kurtosis
N
5.9615
0.8282
0.6859
-1.02716
1.44419
13
M inimum
1st Q uartile
M edian
3rd Q uartile
M aximum
7.0
4.0000
5.5000
6.0000
6.5000
7.0000
5.4611
6.4620
6.5000
0.5939
1.3671
Mean
Median
5.50
5.75
6.00
6.25
6.50
40
Copyright 2013 Tata Consultancy Services limited
22 November 2013
41
Copyright 2013 Tata Consultancy Services limited
22 November 2013
ANOVA: Example
Test for Equal Variances: Stacked versus
Process
Test for Equal Variances for Stacked
Process A
3.25
0.197
Levene's Test
Test Statistic
P-Value
Process
Process
Process A
Process B
Process C
Bartlett's Test
Test Statistic
P-Value
1.84
0.173
Process B
Process C
Since p value > 0.05 through Bartletts test, data passes the
test of equal variances assumption.
42
Copyright 2013 Tata Consultancy Services limited
22 November 2013
ANOVA: Example
Stat > ANOVA > One way (Unstacked)..
43
Copyright 2013 Tata Consultancy Services limited
22 November 2013
ANOVA: Example
One-way ANOVA: Process A, Process B, Process C
Source DF
SS
MS
F
P
Factor 2 12.043 6.022 8.12 0.001
Error 36 26.686 0.741
Total 38 38.729
S = 0.8610 R-Sq = 31.10% R-Sq(adj) = 27.27%
Individual 95% CIs For Mean Based on
Pooled StDev
Level
N Mean StDev ----+---------+---------+---------+----Process A 13 5.0231 1.0709 (-------*-------)
Process B 13 5.9615 0.8282
(-------*-------)
Process C 13 6.3462 0.6253
(-------*-------)
----+---------+---------+---------+---4.80
5.40
6.00
6.60
Pooled StDev = 0.8610
Interpretation:
Since p < 0.05, the difference in the response time is significant and the process
can be called a significant factor.
44
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Proportion Testing
Proportion Testing is used to understand whether the proportion
created by the factor level is significant.
It can be of 2 types:
Ho : PA
Ha : PA
>=<
P0
Ho : PA
P0
Ha : PA
>=<
PB
PB
45
Copyright 2013 Tata Consultancy Services limited
22 November 2013
46
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Issues
Complaints
Resolved %
25
35
30
36
32
33
34
36
28
30
29
32
31
28
35
25
35
30
36
32
33
34
36
28
30
29
32
31
28
35
Data Suggests:
Total no. of trials: 30
No. of events of complaints resolved >= 30% : 22
Ho : PA
0.8
Ha : PA
>
0.8
47
Copyright 2013 Tata Consultancy Services limited
22 November 2013
48
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Sample
1
Interpretation:
Since p > 0.05 through 1 P test, it is not advisable to say that
the team is resolving at least 30% of complaints per day more
than 80% of the times. Process managers claim of providing
resolution on more than 80% of the occasions is not valid.
49
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Data Suggests:
Team Member A:
Total no. of invoices resolved: 60
Total no. of invoices without error: 32
Team Member B:
Total no. of invoices resolved: 65
Total no. of invoices without error: 48
50
Copyright 2013 Tata Consultancy Services limited
22 November 2013
X N Sample p
32 60 0.533333
48 65 0.738462
Interpretation:
Since p < 0.05 through 2 P test, the performance of Team Member A can
be considered significantly less than performance of team member B.
Hence process manager can select member B for promotion.
51
Copyright 2013 Tata Consultancy Services limited
22 November 2013
52
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Contingency Table
Contingency table is used when both output and input variables are
attribute in nature. It uses Chi square test for reaching to the
conclusion.
53
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Hypothesis:
Ho : Hiring of a person is independent of his/ her age
Ha : Hiring of a person is not independent of his/ her
age
54
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Total
Hired
Not Hired
Old
30
150
180
Young
45
230
275
Total
75
380
455
22 November 2013
56
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Total
Total
180
45
45.33
230
229.67
275
75
380
455
Chi-Sq =
0.004 + 0.001 +
0.002 + 0.000 = 0.007
DF = 1, P-Value = 0.932
Interpretation:
Since p > 0.05 , the hiring of a candidate is not dependent upon his/ her age.
57
Copyright 2013 Tata Consultancy Services limited
22 November 2013
Hypothesis:
Ho : There is no relationship between gender & dexterity
Ha : There is a relationship between gender & dexterity
58
Copyright 2013 Tata Consultancy Services limited
22 November 2013
58
59
Copyright 2013 Tata Consultancy Services limited
22 November 2013