SIX SIGMA ASSIGNMENT

XAVIER INSTITUTE OF MANAGEMENT, BHUBANESWAR

Submitted By: Group 10

U113062 Abhinab Mitra
U113063 Abhinav Prakash
U113080 Chinmay Mohapatra

PART - A
Q1. Discuss in brief the Root Cause Analysis methodology during execution of a Six Sigma
Project.
Root Cause Analysis is a process or procedure that helps guide people to discover and
understand the initiating causes of a problem, with the goal of determining missing or
inadequately applied controls that will prevent recurrence. Root causes are underlying, are
reasonably identifiable, can be controlled by management and allow for the generation of
recommendations.
Various methods and tools in RCA:
1. Apollo Root Cause Analysis: It is a 4-stage process which facilitates a thorough
incident investigation. It makes problem solving easy and gives those who have been
trained the skills to solve real world problems faster, more efficiently and effectively
at any time. The logical and practical nature of this method is supported by the
software tool - RealityCharting.

2.

Events and Causal Factor Charting: This is a complicated process that first
identifies a sequence of events and aligns them with the conditions that caused them
along a time line. Events and conditions that have evidence are shown in a solid line
but evidence is not listed; all other observations are shown in dashed lines. Then an
assessment is made by "walking" the chart and asking if the problem would be
different if the events or conditions were changed. This leads to identifying the causal
factors which are identified by evaluating a tree diagram.
3. Change Analysis: This is a six-step process that describes the event or problem, then
describes the same situation without the problem, compares the two situations,
documents all the differences, analyzes the differences, and identifies the
consequences of the differences. Change Analysis is a very good tool to help
determine specific causes or causal elements, but it does not provide a clear
understanding of the causal relationships of a given event.
4. Barrier Analysis: These incident analyses identifies barriers used to protect a target
from harm and analyzes the event to see if the barriers were held, failed, or were
compromised in some way by tracing the path to the threat from the harmful action to
the target. The cause of this failure is then identified as the root cause. Barrier analysis
can provide an excellent tool for determining where to start your root cause analysis,
but it is not a method for finding effective solutions because it does not identify why a
barrier failed or was missing.
5. Tree Diagrams: This type of root cause analysis is very common and goes by many
names such as Ishikawa Fishbone Diagram, Management Oversight and Risk Tree
(MORT) Analysis, Human Performance Evaluation Systems (HPES), and many other
commercial brands. These methods use a predefined list of causal factors arranged
like a fault tree with categories and sub-categories. All categorization methods use the
same basic logic with the premise that every problem has causes that lie within a
predefined set of categories. These methods ask you to focus on one of the categories

6.

7.

8.

9.

such as people and, in reviewing what you know of your event, to choose some causal
factors from the list provided.
Why-Why Chart: It is the most simplistic root cause analysis process and involves
repeatedly asking why at least five times or until you can no longer answer the
question. This method produces a linear set of causal relationships and uses the
experience of the problem owner to determine the root cause and corresponding
solutions.
Pareto Analysis: This is a statistical approach to problem solving that uses a database
of problems to identify the number of predefined causal factors that have occurred in
your business or system. It is based on the Pareto principle, also known as the 80-20
rule, which presumes that 80% of your problems are caused by 20% of the causes.It is
intended to direct resources toward the most common causes.
Fault Tree Analysis: Fault Tree Analysis (FTA) is a quantitative causal diagram used
to identify possible failures in a system and works well to identify possible causal
relationships. It requires the use of specific data regarding known failure rates of
components. Causal relationships can be identified with "and" and "or" relationships
or various combinations thereof. It is often used to support an RCA.
Failure Modes and Effect Analysis: Failure modes and effects analysis (FMEA) is
similar to fault tree analysis in that it is primarily used in the design of engineered
systems rather than root cause analysis. Like the name implies, it identifies a
component, subjectively lists all the possible failures (modes) that could happen, and
then makes an assessment of the consequences (effect) of each failure. Sometimes a
relative score is given to how critical the failure mode is to the operability of the
system or component.

Example: Fishbone diagram showing the root causes of why lean production is not more
prevalent today despite an abundance of favorable attributes.

PART - B
Q1. The data in worksheet MPG contains information on 11 variables for n = 93 makes
and models of automobiles. It is desired to construct a model to predict MPG Highway. 10
possible predictor variables were considered. They are 1. Horsepower, 2.Fuel tank,
3.Passengers, 4.Length, 5.Wheelbase, 6.Width, 7.U-Turn Space, 8.Rear seat, 9.Luggage
and 10.Weight. Fit a parsimonious model, i.e., a model involving as few variables as
possible, provided none of the omitted variables would add significantly to the predictive
capability of the model.
Solution:
To predict the behaviour of a variable dependent on a number of independent variables a
multivariate regression model can be used to find the degree of significance of the
independent variables on the MPG and reduce the number of variables needed subsequently.
The output when all the variables are considered is:

The R square value of 0.746 (74.6%) indicates that the model fits the data very well.
Moreover, the very low P value (0.000), which is less than 0.05 shows that some of the

independent variables being studied here are definitely significant. Looking at the table, we
eliminate Length, Width, U turn space, Rear Seat, Luggage and Horsepower as they
have very high P values.
The modified model with four variables only i.e Weight, Fuel Tank, Passengers and
Wheelbase, gives the following result:

The residuals and the observation order have been plotted on the X axis of Graphs 3 and 4
respectively.

Thus, from the second output (with R square = 0.726 (72.6 %) and p =0.00) the final
parsimonious model is:
MPG Highway=24.1 ( 0.439Fueltank ) ( 0.00834Weight ) (1.01Passengers )+(0.414Wheelbase)

Q4. The data on time to resolve customer complaints in Hours are collected.
Management has put a target for resolving Customer Complaint as max 96. Calculate the
Sigma Rating of the process. (worksheet: TAT)
Solution:
Calculation
(The chart below shows the data and the calculations)
Target Max

96

Resolution
Time

Deviation

Defects/Non-Defects
1= Defect,
0- Not a defect

61

-35

129

33

77

-19

115

19

79

-17

95

-1

88

-8

67

-29

158

62

67

-29

202

106

37

-59

32

-64

86

-10

187

91

72

-24

78

-18

132

36

42

-54

50

-46

90

-6

93

-3

114

18

65

-31

86

-10

30

-66

155

59

94

-2

95

-1

33

-63

50

-46

195

99

73

-23

91

-5

41

-55

99

87

-9

91

-5

34

-62

151

55

103

93

-3

112

16

163

67

78

-18

62

-34

58

-38

107

11

112

16

72

-24

Total

Defects
50

DPO
DPMO
Sigma Rating

16

0.32
32000
0
1.9-2.0

From the SixSigma-DPMO

table
Number of opportunities = 1
Number of defects = 16
Defects per unit = Number of defects/Number of units = 16/50=0.32
Defects per Million units (DPMO) = 1000000* [Number of defects/(Number of units *
Number of opportunities)] = 1000000*[16/(50*1)] =320000
From the DPMO-Six Sigma chart, this value of DPMO (320000) corresponds to a sigma
rating between 1.9 and 2.0.
Hence, we conclude that the sigma rating of the process is between 1.9 and 2.0.