What is Design of Experiments (DOE)

Designed of Experiments (DOE) is a structured approach for varying process and/or product factors (x’s) and quantifying their effects on process outputs (y’s), so that those outputs can be controlled to optimal levels. 

DOE deals with identification of critical factors and their reponse variables, and the magnitude of the response for each level of each of the critical factors. DoE is also used to understand the interaction between the various critical factors to ensure right mix of the critical factors to obtain the optimum amount of response.

DoE is used to understand the transfer function and mathematical model for the optimization of the response variable.

DC motor manufacturer might wish to understand the effects of two process variables, wire tension and trickle resin volume, on motor life. In this case, a simple two factor (wire tension and trickle resin volume), two level (low and high values established for each of the two factors) experiment would be a good starting point. Randomizing the order of trials in an experiment can help prevent false conclusions when other significant variables, not known to the experimenter, affect the results. There are a number of statistical tools available for planning and analyzing designed experiments.

Sixth principle of SPC - Variation due to assignable causes tend to distort the Normal Distribution Curve.

A frequency distribution is a tally of measurements that shows the number of times the measurement is included int he tally. From this frequency distribution we can see if there are only chance causes present in the process of any assignable causes are acting.
If there is a distortion from the normal curve, we can say that there is presence of assignable causes. This finding can actually help us to find the causes and address them.
Various effects of the presence of assignable causes, will tend to distort the shape in center, or the spread as sees earlier. 

Fifth Principle of SPC - It is possible to determine the shape of the distribution curve for the parts produced by any process.

We can learn abut what the process is doing, against what we want the process to do. For this we need to measure the output of the process with the design specifications.the process can be altered if we donot like the comparison, especially if we see a variation.
We need to address eh variation so that it falls in the required pattern. The variation is due to mainly of 2 types. Common Cause variation and Special Cause Variation.
If the variation in output is caused only by common causes, the output will vary in a normal and predictable manner. In such cases, the process is said to be "stable" or "in a state of Statistical Control".  While the individual measurements may differ from each other, they tend to follow a Normal Distribution.
The normal distribution is characterized by the following

1. Location (Typical Value)
2. Spread - Amount by which the smaller values differ from the center.

The shape of the distribution will deviate from the normal curve in case of any un usual occourances.  These changes can be called as Assignable causes.

The presence of assignable causes will result in difference from the usual normal curve, either in Shape, or in spread or a combination of both. 

Non Normal

some changes are given below. 

Normal

Non Normal


The above findings will lead us to the sixth principle of SPC - Variation due to assignable causes tend to distort the normal distribution curve.

Principles of SPC - Variation in a process or Product can be measured

We have already discussed about the same thing done by us giving different output in the first principle of SPC.
Some Variation is always inherent to our job and this is acceptable to some extent so far as the variation is within the Tolerance. However, the Variation tends to increase over a period of time. We need to measure and monitor our job to see that the variation is well within the normal expectations. If we donot make an effort to do so, we land up in trouble and the consequences add to the costs.
Even though it is always desirable to Measure the output of a process, it becomes necessary to measure the output of the process or operation to know when the trouble is brewing.
The measurements can be on the characteristics of the output. It can be the Continuous Variables dimensions, or attribute Variables like colour, shape, finish etc.
After collecting the information as described above, we must analyse to see if things are OK. When we check the output of the feature,  we will quickly notice a Feature. This feature noticed is the basis of third principle of Statistical Process control  - Things Vary according to a definite pattern 

Principles of SPC - No two things are exactly alike.

From the past experience of many generations,  we can clearly understand that things are never exactly alike. All you can find is two similar things. Even the "peas of pod" which look alike, may show some differences among them when we have a closer look. The peas are different in size, shape, colour, softness, or some thing else.
If we apply this to a product, say some products, parts or components, we know that no two manufactured parts are exactly alike each other.  In one way or the other, these parts will be slightly different in Size, Shape, or finish.
If two parts are looking alike, the differences can be found if the resolution of the measurement. The more precise you are in measuring, the differences are more clearly understood.
This is a basic problem, which will get us into trouble for making parts interchangeable, which the main aim of mass production. To work around the problem, we use tolerances.

However, our aim is to keep the variation between the parts to be minimum and as small as possible.
The above discussion is the first principle of SPC. Based on the above discussion, we can go to the second principle of Statistical Process Control (SPC) - Variation in a process or product can be measured.

Six Basic Principles that are foundation for Statistical Process Control

The six principles below are the foundation for Statistical process Control. These can be clearly understood using Frequency Distributions.
The principles are listed below. The explanation is linked to each sentence

1. No two things are exactly alike
2. Variation in a product or process can be measured
3. Things Vary according to a definite pattern
4. Whenever things of the same kind are measured, a large group of the measurements will tend to cluster around the middle.
5. It is possible to determine the shape of the distribution curve for the measured Output (Parts produced, transaction) by any process
6. Variation is due to assignable causes tend to distort the Normal distribution Curve.

To Understand more on these principles, we can study the data from the output of the process. These foundation principles will be useful for all types of processes.

Roles and titles in a Six Sigma project

Black Belt 
Leaders of team responsible for measuring, analyzing, improving and controlling key processes that influence customer satisfaction and/or productivity growth. Black Belts are full-time positions.


Green Belt 
Similar to Black Belt but not a full-time position.

Master Black Belt  
First and foremost teachers. They also review and mentor Black Belts. Selection criteria for Master Black Belts are quantitative skills and the ability to teach and mentor. Master Black Belts are full-time positions.


Customer 
Any internal or external person/organization who receives the output (product or service) of the process; understanding the impact of the process on both internal and external customers is key to process management.


Process Owner

Process owners are exactly as the name sounds - they are the responsible individuals for a specific process. For instance, in the legal department there is usually one person in charge - maybe the VP of Legal - that's the process owner. There may be a Director of Marketing at your property - that's the process owner for marketing, and for the Check-in process, the process owner is typically the Front Office Manager.

Team leader

For DMAIC projects, the team leader is usually the Black Belt. For Quick Hit and iDMAIC projects, it is typically the Sponsor or Process Owner. For large DMAIC projects with more than one BB or MBB, the Team leader is the main point of contact for the project.

Team member

An active member of a Six Sigma Project team (DMAIC or iDMAIC),heavily involved in the measurement, analysis and improvement of a process. To be effective, team memberships require a minimum of 10% time commitment to a phase of the project. He/she also helps fosters the Six Sigma culture within the organization by informing /educating fellow Associates about Six Sigma tools and processes.

Transfer Team Leader (Process Owner/Department Head)

A person selected by the GM and property SIXSIGMA Council to lead an iDMAIC project based primarily on proximity and decision-making authority relative to the process involved. This person has primary responsibility for implementing the project, leading the team, and interacting with others to gather information and understanding necessary to succeed. Often, the transfer team leader will be the department head or process owner of the process being improved with the best practice. The ability to lead the team and to anticipate clear barriers are important characteristics for a person in this role.

Transfer Team Member

Associates selected by the Transfer Team Leader and Six Sigma Council to serve on the iDMAIC project based on their knowledge of key aspects of the process, experience with the current process, enthusiasm for improvement, and ability to champion change. Other key factors in selecting transfer team members include time availability and representation from relevant functions. All members will be provided training on the skills and tools used in the transfer process.

Project Sponsor

This member of the executive committee is a strong advocate of the project and can assist with barriers that may come up. He or she is accountable for the project's success and can therefore explain to Six Sigma Council members and everyone in the property the business rationale for the transfer project and assist with cross-functional collaboration efforts. He or she will remain up to date on key aspects of the project by regularly meeting with the team leader and members.

The project sponsor:

• Is a member of the Executive committee
• Is accountable for project success
• Addresses cross-functional or other barriers
• Reviews and tracks progress with team leader
• Advocates for necessary resources

Statistics -3 - Presentation/Organization of Data - Tabulation

The data collected by using various methods like Surveys, Interviews, In-field studies etc will give you the raw data. You may not be able to draw any conclusions using this data. This data need to be organized and then only it can talk(yes. the data talks if it is collected properly) to you. Also this data will talk for you only if your present it in such a way that the user can receive.
There are various methods by which you can organize and present the data.The selection of the method depends upon the purpose and the target audience.
The most commonly used method is tabulation.  Normally we see many tables daily. However, if your presentation or analysis is to be relevant, the table shall contain the relevent details.
A good table is used to condense the data and present in a useful form. It is the most common method and easily understood method of presenting data.


A good table will have the following details.

1. The Table must have a heading.
2. The Table should present the data clearly, highlighting important details.
3. The Table should save space but attractively designed.
4. The table number and title of the table should be given.
5. Row and Column headings must explain the figures therein.
6. Averages or percentages used in the table should be close to the data.
7. Units of the measurement should be clearly stated along the titles or headings.
8. Abbreviations and symbols should be avoided as far as possible.Sources of the data should be given at the bottom of the data.
9. In case irregularities creep in table or any feature is not sufficiently explained, references and foot notes must be given.The rounding of figures should be unbiased.
10. Wherever notes is required, they should be given below the table with relevent references.

If a table contains all the above, it can be said a good table. Eventhough these are present is the normal table, the classifications is good to verify if the table is good or not.

Statistics -2 - Data Collection - Types of Data

The first step of any statistical enquiry is the collection of relevant numerical data. The data used for statistical purposes is mainly of two types.

Primary Data - Data collected for the purpose of the given inquiry is called as Primary Data. These are collected by the enquirer, either by his own or through some agency set up for this purpose, directly from the field of enquiry. This type of data can be used with greater confidence because the enquirer himself decides upon the coverage of the data, definitions to be used and as such will have a greater control over the reliability of the data.

Secondary Data  - The data already collected by some other agency or for some other purpose and available in published or un published form is known as secondary data. The user has to be perticularly careful about using using such data. The user must clearly understand the nature of the data, their coverage, the definitions used for the data and their reliability.

The usage of secondary data is generally preferred if the conditions mentioned above are clear and usable. This will reduce the time taken for the analysis, also reduces cost of the analysis.

usage of Sampling -  The big question is weather the collection of data should be done by complete population or by sampling. If sample is used, care should be taken that this is a representative of complete population. A sample designed with care can produce results that may be sufficiently accurate for the purpose of enquiry. A Carefully designed sample can save a lot of time and money. 

Methods of Data Collection : The methods used to collect data are Questionnaire Method, Interview Method and Direct Observation Method. Any one or a combination of these are used to collect data.
Usage of Data :  The data collected should be subjected to a thorough scrutiny to see if they may be considered correct. The success of the analysis depends on the reliability of the data. However excellent the statistical method of data analysis may be, they cannot bring out useful and reliable information from faulty, unreliable of mistaken data. Especially, this is more applicable in case of usage of secondery data.

Like this?? - Go on the visit the next column - Statistics - 3: presntations and Organization

Statistics - 1 - What is Statistics ?

Statistics can be described as a quantitative method of scientific investigations.

If used as  plural noun 'Statistics' means the numerical data arising out of any sphere of human experience.

Used as singular 'Statistics' is the name for the body of scientific methods used for collection, analysis, Organizing, and interpretation of Numerical data.

According to American Statistical Association 'Statistics is the scientific application of mathematical principles to the collection, analysis, and presentation of numerical data'

Also, There is a different meaning for the word 'Statistic' in the field of Statistics(subject). In this sense A 'Statistic' is a numerical item which are produced by the some calculations using the data. Standard Deviation, Mean etc are called as 'Statistic'  in this sense.

This is one arm of Mathematics, which is extensively used in all most every field. Statistics has become an important tool in the work of many academic disciplines such as medicine, psychology, education, sociology, engineering and physics, just to name a few. Statistics is also important in many aspects of society such as business, industry and government. Because of the increasing use of statistics in so many areas of our lives, it has become very desirable to understand and practice statistical thinking. This is important even if you do not use statistical methods directly.

Even with so many uses, there is some mistrust in public about statistics. This is because of the misuse of the figures by the people for their convenience. It is often said by people that statistics can prove any thing. During the introduction to the course i joined on statistics, this statement is used. There are 3 types of statistics. 1 - Lies, 2- damned Lies 3- Statistics. We will teach you the 3rd part here.

Used properly statistics is a panacea for all the problems faced by the world. it can be a tremendous tool for the growth of any organization. 
Visit the next post Data Collection - Types of Data