Pradeep Shukla

Cause and Effect Diagram
 
This matrix is generally known as a Fishbone diagram or Ishikawa diagram.
It is a visual tool which is used to analyze and display the related potential causes of a specific problem.
This diagram looks like skeleton of a fish.
 

 
Typically, it has structure around the main branches of the Fishbone. Which are here below.
 
·       People
·       Process
·       Equipment
·       Materials
·       Environment
 
When we generally visually mapping out these potential causes and their relationship in diagram format. It usually identifies the main root cause of the problem.
This is a great tool in problem-solving and process improvement.
 
 
Cause-Effect Matrix
 
This is generally a Six -Sigma valuable tool which is used to determine the Key Process Input Variables which are based on priorities of customer.
It is also known as X-Y diagram, prioritization Matrix and Correlation Matrix.
The main goal of this matrix is to mathematically compute the relationship between Key process input variable and Customer output.
 
Prerequisite of this matrix:
 
·       Process Map
·       Voice of customer
·       Cause and effect analysis
 

 
·       It generally relates process steps to input and correlates to process output.
·       Customer requirements are ranked by order of importance and then inputs and outputs are rated by their interaction impact.
·       It identifies key customer requirement.
·       Outputs are generally given a priority scores.
 
Inputs are rated based on the strength of their relationship with output variables as per below.
 
0 = No correlation
1 = Remote correlation
3 = Moderate correlation
9 = Strong correlation
 
Few major differences between Cause -Effect Matrix & Cause and effect Diagram are here below:
 
 
Cause-Effect Matrix
Cause and Effect Diagram
Purpose
Primarily used to prioritize and quantify the potential cause of a problem
Used to visually brainstorm and categorize potential causes of a problem
Format
It is generally a matrix or table form which lists the potential causes in rows and criteria in a column.
It is a visual diagram that looks like a fish skeleton. A horizontal line representing the problem or effect and “bones” branching out to represent various categories of potential causes.
Analysis Method
It involves in quantitative analysis. It assigns numeric values
It does not assign any numeric values. It focuses on qualitative analysis and creative thinking.
Focus
It supports team to make data driven decision making.
This is a tool which generating ideas and promoting a better understanding about problem.
 
 
 

Multi Generational Product Plan
 
It is generally a strategic roadmap that designs the development and evolution of a product compare with multi-generation.
This is generally including a long term vision for the product.
 
Key Elements :-
Product Vision – Vision should be very clear to achieve the aim of problem to be solve.
Roadmap – A clear timeline and outline of the plan should be release of the product
Market analysis – A good market research should be conducted to identify customer needs, trends and opportunities.
Technology Considerations – A good technology stack and infrastructure is required for product growth.
Resource allocation – A rich allocation of resources like – budget and time.
Risk assessment – Timely identification of potential challenges and risk involved and make plan to mitigate them.
Customer Feedback – A regress feedback from customer is required.
 
Few real-world examples –
 
Apple iPhone – Started from 2007 it has drastically updated the product as per new generation.
Microsoft Windows - Microsoft Windows has started in 1985 with Window 1.0 to Window 11. Each one introduced with improved features as per customer requirement.
Tesla Electric Vehicles – It also started with basic models and upgraded to new models based on new demanding features and performance.
Video Game Consoles – These are very popular in kids. Products like play station started with PS1 to PS5 based on new generation with powerful hardware.
Other smartphones – Smartphones has drastically upgraded as per customer requirements. Started with basic memory in KB to memories in GB with introduce of multi features which are required as per new generations.
 
Major differences in Multi-Generational Product Plan (MGPP) and DMADV
 
 
MGPP
DMADV
Purpose
Primary used for product development and improvement.
Used for process improvement and redesign.
Phases
Involves phased approach
Involved structured 5 phase approach
Focus
Doesn’t follow a strict set of phases
Follow strict set of 5 phase approach
Approach
Starts with product and aim to enhance it over time
Starts with a problem or quality issue in mind
Iterative Approach
Both involves in iterative approach
Both involves in iterative approach
Continuous Improvement
Both methodologies are aimed to achieve continuous improvement
Both methodologies are aimed to achieve continuous improvement
 

Matrix Diagram
 
It is generally a visual representation diagram which is used to categorize, organize, and inspect any relationship between multiple items.
This is a grid like structure, in which rows and columns shows different and multiple categories or factors.
Generally, the intersections between these rows and columns we filled with any symbols, colors or any kind of value which can show the presence, strength of relationship between the items.
This diagram generally used in various department like – project management, quality control, decision- making.
 
Below are the various types of Matrix Diagram and their usage:
 
 
Usage
Example
L Shaped Matrix
Used in analyzing the relationship two sets of factors
Manufacturing industries using this type of matrix to know the most suitable combinations
Y shaped Matrix
Used to analyzed 3 sets of factors
Generally used in Marketing to know the compatibility of targets
C Shaped Matrix
Used to analyzed 3 sets of factors where one set can influence the other set
Used in Healthcare department to know how a medical treatment can impact the patient
X shaped Matrix
Used to analyze and help establish relationship between customer requirement and product features.
Used in Six Sigma and quality management.
Also, an automotive industry can use this matrix to know the relationship between customer desire and and product feature (fuel efficiency).
Interrelationship Matrix
Used to understand strength and direction among factors
Product management teams are using to visualize the different tasks involved.
Prioritization Matrix
Used in decision making by rankings based on criteria.
Can used in multiple type of industries where they want to analyze and want to make decision for new location, new store etc.
 
In the end, all types of matrix are very useful and helpful for different type of work and requirement. Choice of matrix depends on situation and type of problems and relationship we want to analyze.

Box-Behnken Designs
 
This design was explored by statisticians George E. P. Box and Donald R. Behnken. So, it gave name of Box-Behnken design.
It is a type of experimental statistical design which is generally used in statistics and engineering to improve the process and conducting experiments.
This is used to find the optimal combination of input variables which can lead to achieve desired output.
This design is generally useful when we want to know the relationship between variables. This is used when there is quadratic, and interactions are considered among variables.
Whereas a full factorial design generally consists of all possible combinations which are available. In this test we can study the effect of each factor.
 
Below are the common differences:
 
Box-Behnken Design
Full Factorial Design
No of experiments
This is related to response surface methodology, so uses fewer experiments
In this method we can test all possible combinations for each factor.
Efficiency
More efficient with fewer experiments
More efficient with comprehensive experiments. It increases the experiments with the no of factors and levels.
Complexity of analysis
Simpler in comparison
Complex when dealing with large numbers
Application
Useful when no of experiments are limited.
More useful when no of experiments are more and have complex relationship
 
 
 
 
Advantages and disadvantages of Box-Behnken Design:
 
Advantages
Disadvantages
Box-Behnken is more efficient with full factorial when fewer experiments are allowed with limited resources
It has limitations in experiments
It is less complex that full factorial design
Generally used selective sample for experiments, so sometimes missed optimal points
Helps to identify optimal factor level
It is very less flexible, so sometime results may not be accurate
This design allows for interpolation
It is not providing a comprehensive analysis as compared to full factorial designs
Can distribute factor level evenly
More focused on quadratic relationship, can ignore linear effects.
 

1 Sample Test
 
A one sample test is a hypothesis test which generally used to determine whether our sample’s characteristics are significantly stands different from a known or hypothesized parameter.
Also, it generally used for single population group or sample.
 
There are 2 most common tests are available which used to compare the sample:
·       1 sample T-test
·       1 sample z-Test
 
 
1 sample T-test
1 sample z-Test
Sample size
It is more robust for sample size <30
More suitable for large sample size, sample size >30
Population standard deviation
Used when population standard deviation is unknown
Used when population standard deviation is known
Application
Used when we want to determine whether a sample mean significantly differ from a hypothesized mean or sample in small sample size
Used when we want to determine whether a sample mean significantly differ from a hypothesized mean or sample in large sample size
Usage
We use when we generally want to compare a sample mean with population mean
We generally used sample standard deviation instead of population standard deviation.
Assumption
Both tests assume data is normally distributed
Applicability
More relevant as it can handle unknown population standard deviation with small sample size
Relevant only if we have large sample size with known standard deviation
Robustness
More robust with small sample size
Less robust with small sample. Used only for large sample
Calculation Method
Involves degree of freedom
No degree of freedom adjusted
Software Availability
Available with all the statistical software
Less available with all
 
Below are some real-world examples:
1 sample T-test – Let suppose a battery company want to check the charging life of newly battery which just developed. So, company will select some battery randomly and check for their usage time.  After getting usage time they will check whether it is significantly different from population mean.
 
1 sample Z-Test – Let suppose a researcher want to check whether a newly developed drug can lower the blood pressure or not. So, take a known mean which is 120 mmHg. So, researcher will take sample which would be generally more than 30 to determine whether this drug can lower the blood pressure with known mean.

Mann Whitney test
 
This test is known as Wilcoxon rank-sum test.
It is a non-parametric statistical test which is generally used to review the distributions between two independent groups.
This is also an alternative to T-test, but we should use this test when our assumptions of normality and equal variances generally not met. This test is also applied to check that medians of two groups are statistically different from each other or not?
 
This test works on below parameters:
·       Ranking
·       Calculating U
·       Comparison      
·       Hypothesis testing
·       Interpretations
 
 
Mann Whitney
T test
Assumptions
No specific distribution assumptions
Applied when data is non-normal and skewed.
Comparing Rank between 2 groups
Assumes data follows a normal distribution.
Comparing with 2 groups with equal variances
 
 
 
 
Real scenario where Mann-Whitney U test accepts more than T-test.
ü  When data is ordinal
ü  When we have small samples
ü  When we have outliers
ü  When our data is skewed
ü  When variances are not equal
ü  When we have categorical data to compare
 

1 Proportion Test
 
1 Proportion Test is generally known as one-sample z test for proportions.
This is a statistical method which commonly used to determine that the proportion in certain characteristics in our used sample is significantly different or not from our hypothesized proportion.
This test is generally helpful in single group.
 
As we know that this is a statistical tool so we generally use this tool in Measure and Analyze phase in DMAIC methodology.
 
Measure - In Measure phase, we collect the related data to understand the current situation of our problem. Also, we try to understand the problem with data which we are trying to investigate.
So, we use this 1 Proportion Test in measure phase to determine that we are assessing that whether a proportion which we are measuring is significantly different or not? Like – defect rate etc.
 
Analyze – In this phase, we generally analyze our problem with the help of our data to identify what are the root causes for the said problem. Also, we want to identify the relationship between cause and problem.
So, it generally helps researchers to identify if a certain proportion is deviating and if this statistically significant or not?
For example, if we are working on improving customer satisfaction level and we believe that the actual proportion of satisfied customer is not meeting our target. So, we can apply this test to know whether proportion of satisfied customer is significantly different from our said target or not?
This test is generally more helpful when we have data, and we want to do decision making on the basis of said data.