iSixSigma

Whitehurst

Forum Replies Created

Forum Replies Created

Viewing 100 posts - 1 through 100 (of 285 total)
  • Author
    Posts
  • #186754

    Whitehurst
    Participant

    The basic underlying assumption is a Normal distribution, which is the basis for all SPC.  Any underlying distribution that is averaged in sample sizes of 5 becomes normally distributed.

    0
    #186753

    Whitehurst
    Participant

    The rules were develooped by someone at Western Electric, thus the name, just like the majority of statistical manufacturing techniques were developed at Bell labs, the student t is the prime example.

    0
    #186661

    Whitehurst
    Participant

    You can try an online certification if you’re fine with it. Else, i don’t know any LSS school within that area.

    0
    #65341

    Whitehurst
    Participant

    Six Sigma is really what Demming intended quality engineering to be.  8D is not outdated as several institutions use this report format to display corrective action, same as 7 step, CAR’s or any other customer required root cause analysis tool.  Most orginizations do not have the time or resources to conduct a six sigma project which is a formal drawn out approach to an 8D.

    0
    #65190

    Whitehurst
    Participant

    In addition to what Dave had to say, LSS can be used to initially map whatever processes exist, then when implementing a new PA, LSS tools may be used to make the new processes more robust, with minimal NVA – DFSS, where the “Design” applies to the design of the process, rather than a product.Once processes are established, built on the CMMI framework, LSS tools are ideally suited for improving upon processes.

    0
    #176848

    Whitehurst
    Participant

    Can I also get a copy ofthe dot game- This will be an excellent tool to help our participants in our Kazien events. e-mail [email protected] – thank you

    0
    #65184

    Whitehurst
    Participant

    The business I work for is just beginning CMMI implementation.The way we are approaching it is initially all development sites receive a benchmark appraisal
    (we’re at Level 1 – no surprise there)The appraisers provide a detailed report including status of all PA’s: SG’s, GG’s,Sp’s etc.At this point, we begin planning activities to bring our processes to a point where we will meet the requirements for Level 2.We are using a hybrid between the staged and continuous representations. Once we are appraised as a Level 2 business, we will begin working from the Continuous representation, focusing on PA’s that will provide the most benefit first, then working on other PA’s.One thing that is EXTREMELY important is to have CMMI implementation be a company directive. Unless it has executive endorsement, you will find that it will be next to impossible to implement.

    0
    #176324

    Whitehurst
    Participant

    In regards to query three special causes is a source of variation that causes a fundamental change in a process. Special causes distort a normal distribution.

    0
    #176223

    Whitehurst
    Participant

    Thanks for your help.

    0
    #176124

    Whitehurst
    Participant

    Hello,
    Thanks for this, however I would also like to know the overall approach for doing a capability trial in terms of assessing whether a new piece of equipment was suitable to use for the manufacture of product.  Does anyone have instructions on how to go about this?
     

    0
    #65076

    Whitehurst
    Participant

    Hi Carlos,
    I’ve used SS in the area of mainframe networking, which yielded extremely positive results (in excess of $10+M/year at my former employer).  I suggest using multiple MBF approach to address the various opportunities your going after…

    0
    #171292

    Whitehurst
    Participant

    Thanks for this Erik

    0
    #171278

    Whitehurst
    Participant

    Doh – forgot my address for the calculator – thanks
     
    [email protected]

    0
    #171270

    Whitehurst
    Participant

    Hello – could you forward yout Takt calc please.
    cheers

    0
    #171268

    Whitehurst
    Participant

    That is part of what I am trying to understand.  If I first do a two sided test and the result is significant, showing there is a difference. Then based on the results, I do a one sided test, do both one and two sided results always come out significant or not significant providing the correct Ha direction has been set for the one sided tested?
    Hope this makes sense!
    Joe

    0
    #169536

    Whitehurst
    Participant

    Thanks BC – I knew it was supposed to be out of control but I didnt know the control limits were based on the measurement variation – All becomes clear!
    cheers

    0
    #165903

    Whitehurst
    Participant

    I reccomend taking at least one semester of statistics if you want to be a good black belt.  Neither a lack of training in statistics, nor a six weeks seminar in six sigma will make you a “good” black belt.  It not a matter of running Minitab or calculating a hand full of formulas, you have to understand the theoretical bases of the statistics and choose the right formulas and evaluate the computed statistics correctly.  Be cautious with Minitab…it gives you what you ask for.  If you ask it for the wrong thing or don’t know how to interpret the results, Minitab won’t tell you so.

    0
    #165902

    Whitehurst
    Participant

    Dr. Donald Wheeling points out in his book “Introduction to Statistical Process Control” that a distribution as badly sckewed as a right trianagular distribution has little affect on the accuracy or usefulness of making predictions based on the data you computed the statistics from.  However, you can not simply ignor non normal distributions and say that they are normal.  You have to make a decision on how critical the application is.  I would not want to ride on an airplane whose capability was based on left handed skewed poisson distribution of data representing the failure rate ball bearings failing.  Many software packages are now applying curve fitting and calculating CpK based on the number of standard deviations the mean is from the upper process limit.  You can also do curve fitting and convert the PPM to a CpK. 

    0
    #165901

    Whitehurst
    Participant

    There is one more test for means.  At 30 samples or more you should use a Z test.  Blow 30 samples use the “Student’s t test”.  At 6 samples or below you would want to use the “Protected t test” which compensates for the error in small sample sizes.  In all these formulas you should be reasonably sure the data if from a normal distribution or they don’t apply.

    0
    #161147

    Whitehurst
    Participant

    Adam,
      I have never seen the card drop game and would greatly appreciate the information on it.  If you get a chance please forward to:
    [email protected]
    Thanks Much – Joe

    0
    #161066

    Whitehurst
    Participant

    Mutiple Sujective Evaluation can help convert Attribute responses into continuous so that you can do a DOE. The downside is can take a slightly longer time to perform, but will certainly allow you to use the result in a DOE to help you tweak the dials you want to improve your process.

    0
    #160460

    Whitehurst
    Participant

    There is an excellent resource called the Kaizen Desk Reference Standard(R. Vitalo, http://www.vitalentusa.com)It goes thru every step of the process in great detail and should be exactly what you are looking for. I’ve used it since it came out.Here is the direct linkhttp://www.vitalentusa.com/products/kdrs/kdrs_learn_about.phpRegards
    JoeFull Disclosure – I’ve known the author for many years and consider him a friend, but don’t hold that against him!

    0
    #156845

    Whitehurst
    Participant

    Thank you for your help Erik.

    0
    #156744

    Whitehurst
    Participant

    Thanks for your help on this.
    cheers

    0
    #156524

    Whitehurst
    Participant

    SATYA,
     
    Deming principles i understand but WHAT IS A DEMING REPORT ?
    Well can you throw some six sigma light ?
    regards
    joe

    0
    #155632

    Whitehurst
    Participant

    Thanks for all your help, I think I will go with Vidyadhars final explanation.
    Cheers
    Joe

    0
    #155627

    Whitehurst
    Participant

    Would you please send me a pull system calculator?  Thanks.
    Joe

    0
    #155483

    Whitehurst
    Participant

    Well you have not told us the way you are calculating variance
    Minitab calculates variance by using the std ^ 2  but in the numerator it multiplies it with a constant C4, so it looks like c4( n – 1 ) .
    Use minitab if you have it or use excel to calculate it.
    STDEV()
    regards
    joe

    0
    #155482

    Whitehurst
    Participant

    There is a difference between project identification and project selection.
    There is NO cook book  to find the kind of projects thats important to your organization. Find the goal of your organization, try to relate your departmental goal to the organizational goal, try to see which metrics best describes your department and whether improving that departmental metric will improve your organiational objective. Find the key drivers of that metric , collect inital data and see where do you stand and select project if needed accordingly.
    Not so simple as it sounds but good luck.
     
    regards
    joe
     
     

    0
    #154942

    Whitehurst
    Participant

    “WE ARE HERE TO CHANGE, IF WE FAIL CHANGE MANAGEMENT
     
    REGARDS
    JOE

    0
    #154914

    Whitehurst
    Participant

    UCL and LCL are calculated from the data that you have calculated.
    USL and LSL will be your design specifications.
    What do you mean by optimal control limits ? if its UCL and LCL then its calculated  +- 3sigma distance from the mean.
    regards
    joe

    0
    #154716

    Whitehurst
    Participant

    Use Minitab !!
    how ? use minitab help

    0
    #154661

    Whitehurst
    Participant

    Please call yourself MBB not TMBB
     

    0
    #154660

    Whitehurst
    Participant

    Did you the try the minitab help or http://www.google.com

    0
    #154655

    Whitehurst
    Participant

    Quality can be compared to Sex
    Everyone loves it
    CEO’s shy to talk about it
    Every one buys it
    Every ones in for it
    Some are new to it
    Some one wants to learn it
    and some consultants think hey are naturally born with it
     
    Regards
    joe
     

    0
    #154654

    Whitehurst
    Participant

    Quality can be compared to Sex
    Everyone loves it
    CEO’s shy to talk about it
    Every one buys it
    Every ones in for it
    Some are new to it
    Some one wants to learn it
    and some consultants think hey are naturally born with it
     
    Regards
    joe
     

    0
    #154406

    Whitehurst
    Participant

    PLEASE CONTACT MR. KUMAR
    regargds
    joe

    0
    #153684

    Whitehurst
    Participant

    PLEASE STOP RECOGNIZING ASQ CERTIFICATES & SAVE SIX SIGMA.

    0
    #153683

    Whitehurst
    Participant

    Theres a great deal of relation between SS and Innovation. Try Triz-journal.com to learn about innovation and how you can integrate these two things. Also try Syneptics.com.
    Hope this helps!!
     
    joe

    0
    #153625

    Whitehurst
    Participant

    It depends what you are tyring to do
     
    regards
    joe

    0
    #153403

    Whitehurst
    Participant

    The last I counted was 67229 out of which 67228 are in India!

    0
    #152974

    Whitehurst
    Participant

    Productivity  – continous data

    0
    #152893

    Whitehurst
    Participant

    Sir,
    Six sigma can be implemented anywere including south Africa.
     

    0
    #152853

    Whitehurst
    Participant

    USE EXCEL if you are not using

    0
    #152855

    Whitehurst
    Participant

    We are not consultants, we can give you results !!
     

    0
    #152854

    Whitehurst
    Participant

    First of all what are your CTQ’s, there are so many drivers which drives your customer satisfaction a single DMAIC project cannot resolve the problem.
    Secondly, I am sure you are from the service industry . Find a way to seperate customers who are dis satisfied due to call service and dis satisfied due to the product itself.
    Capture reasons why they are dis satisfied?
    Use pareto .
    YOUR OBJECTIVE IS TO INCREASE CUSTOMER SATISFACTION NOT TO USE  HYPOTHESIS TESTING TO IMPRESS YOUR MBB.
    Apply brains, look at the process observe the process, talk to people understand whats happening in the process etc etc.
    Forget hypothesis testing!!!
    joe
     
     
     

    0
    #152847

    Whitehurst
    Participant

    Ha ha ha ha ha ha ha ha ha ha ha

    0
    #152836

    Whitehurst
    Participant

    use normdist excel function to find it.
     

    0
    #152837

    Whitehurst
    Participant

    Can you some explain what is the purpose of MR chart?
    I feel its a chart without any practical usage
    regards
    joe

    0
    #152793

    Whitehurst
    Participant

    I chart

    0
    #152776

    Whitehurst
    Participant

    Hi,
    I have been viewing and listening to many people talking about attrition project specially in service industry. Some claims they found the right x’s and reduced attrition, some has already initiated a project etc. The amount of malpractices in service industry will never allow anyone to reduce attrition.
    I have seen people initiating a project on attrition, have attrited themselves.
    One of the first thing one should do is to list down Demings Laws and see where your organiszation stands with respect to those laws. Just making your organization follow those laws is a project by itself and I am sure it will also take care of attrition plus some other problems also.
    Good Luck !!
     
    regards
    joe
     
     
     
     
     
     
     
     
     

    0
    #152721

    Whitehurst
    Participant

    Hi,
    I have not seen other replies but it seems if you donot take immediate action it will impact your ogranization .
    Do a brainstorming session by inversing the problem :
    What are the things you will do as an organization to make sure that 100% of your customers leaves your organization?
    Find all the reasons and see whether you do it or not.
    If you do it find differnt solutions even the crazy one and run them through solution selection matrix.
    After selecting solutions find ways to find ” how you will make sure that these solutions fail 100%”?
    Plug in the gaps , implement, evaluate results and then work towards ideality ? 
     
    Regards
    joe
     
     
     
     

    0
    #152456

    Whitehurst
    Participant

    Discrimination is the resolution of the measurement system to detect small changes in parts.
    Recommended discrimination : 1/10 of  6 times of the measurement standard deviation.
    ” Number of distinct categories ” will give you this measurement. Generally it has to be more than four or five to have adequate discrimination.
     
    regards
    joe

    0
    #152455

    Whitehurst
    Participant

    Though people will disagree I think both are same there might difference in philosophies, tools , process improment, variation reduction but both are same essence ” a methodology to solve a problem or methadology to reduce variation or many more definitions”
    They are both are organized common sense but unfortunately common sense is not so common.
    regards
    joe

    0
    #151956

    Whitehurst
    Participant
    #151952

    Whitehurst
    Participant

    Thanks for your response .I agree

    0
    #151928

    Whitehurst
    Participant

    Good Morning from Detroit!
    I can feel your pain in regards to understanding your Service Level problems.  Here are a few suggestions:
    First Call Resolution – this metric pinpoints how many calls are handled on the first attempt, ex.. 96% by using this metric, you can understand the effects of repeat calling against the service levels and other metrics.  Increasing the FCR metric is a great project.  You can use a pareto to understand the types of issues that get repeat calls, analyze the issue based on case comments, and implement solutions.  Most importantly you can track your improvements via the Pareto and FCR percentage.  Due to the extreme variability of incoming call flow pattern, it is not very easy to understand the relationships between your variables like average handle time AHT and its attributes.  
    ATT – Average Talk Time
    ACW – After Call work Time
    FU – Follow UP time
    OHT – On Hold Time
    I always enjoy using the above attributes to monitor my variation through the use of Box Plots; you can compare the variations between the attributes which are the causal factors of Handle Times.
    Schedule adherence is an ok number to use, but it really just tells you how close to being on-time or lunch breaks your are, not really going to solve any big problems.
    Another metric that is not going to help too much, but gets a lot of attention is agent productivity.  The AP is usually a measure of Total Staffed seconds, vs. seconds actually working or available to take a call (at your desk, but not on the phone)  The fun thing about this number is that you can cheat it by altering your shrinkage time, which is why I don’t allow to much emphasis on it by my center.
    Taking the VOC approach to the metric is how we usually handle it here.  An overall service level does not really tell you much, especially if it is based on 1/2 hour intervals.  What does your customer actually feel, if 1/2 your intervals are 100% SL and the other 1/2 of the intervals are 0% SL, you are not really doing yourself or more importantly your customers any favors.
    A recent exercise we conducted and will continue to observe is the relationship between 1/2 hour service levels and customer satisfaction, a good link is usually established pretty easily with a simple regression.  We also use a control chart and process capability calculation to understand how they relate, by setting all calls outside of the SL as a defect; you can better understand what is taking place.  In our case we used a U chart (I think) in combination with a Process Cap based on the distribution of the data.  By taking the data out of the continuous range, you can use some pretty useful tools when turning it into an attribute.
    So:
    SL 1/2 hour defects, or by call defects, what is your customer feeling?
    First Call resolution metric increase
    Control chart to understand trends and identify variation issues
    Process Cap to assess the potential of what needs to be done
    Box plots to understand the variation of the different call handling components lined up against each other
    Regression equations comparing customer sat to the different components or SL equations.
    Unfortunately there is no magic bullet, but a combination of the above mention variables should give you a good start.
    Until you get a handle on repeat caller’s variation, via the First Call Resolution metric, you are never going to get a steady forecast.  You may never get one anyway, you should see all the variation we get to try and predict against. 
     
     

    0
    #150533

    Whitehurst
    Participant

    He is an irresistable little darling isn’t he – you were going to leave him and you’ve already given him another 2 replies – maybe there’s some truth in what he’s saying that’s pressing your buttons .

    0
    #147556

    Whitehurst
    Participant

    Theres no function that will draw a normal probability plot in excel.
     
     

    0
    #147052

    Whitehurst
    Participant

    Hi Ashish,
     
    Without understanding the nature of business how can someone give such kind suggestion. Please try to understand the meaning of BPO before giving such kind of stupid suggestions.
     
    regards
    joe

    0
    #146764

    Whitehurst
    Participant

    Nah, nobody could be that stupid. Reigle has gotta be a smart con man.

    0
    #146763

    Whitehurst
    Participant

    Read the previous posts about Motorola sacrificing 60,000 staff … and them and GE sacrificing manufacturing and moving to outsourcing.

    0
    #146665

    Whitehurst
    Participant

    Beautiful example George.
    When will all the SS idiots wake up that you will never know the distribution from which data is taken. You need 3200 data points to check the fit out to just +/-2.95 sigma.
    Forget the SS crap.  You don’t need to know the distribution.  You don’t need transforms.

    0
    #144609

    Whitehurst
    Participant

    Simon,
    Can you send the simulation information to [email protected]?
     
    The teller simulation is one I think anyone can relate too.
     
    Thanks

    0
    #140107

    Whitehurst
    Participant

    I have enrolled for an online certification towards a MSSBB starting with the SSGB program. I  was just wondering if there are reputable  institutions that offers a complete classroom based program.

    0
    #139795

    Whitehurst
    Participant

    I have used SPC XL and found it to be a great package.

    0
    #139343

    Whitehurst
    Participant

    How to put an equation on minitab? I have a large data set, but i cannot put the data in a formula, because I DONT KNOW HOW!!

    0
    #138904

    Whitehurst
    Participant

    If the data is truly non-normal – not because of sampling issues or special cause problems.
    You attempt a Box-Cox transformation, the run the data through the normal distributions
    You may also try and do a best fit to a probability distribution such as weibull, lognormal, or exponentional.  Use a program to do a best fit and pick the one with the highest, I think highest (check), PValue.
    On a side note, the ASQ SSBB is a hard test, things like you just asked are the easier questions on the test.  I have passed that test and it was very challanging. 
    good luck

    0
    #138836

    Whitehurst
    Participant

    For future reference, you may want to designate what AHT stands for.  There are several quality BBs on this forum who can help you that are not call center BBs and are not familiar with the terminology.
    Average Handle Time is a summation of categories such as Call Time(ACD) and other components.
    I would reccomend using your Switch output as a starting point.  Take a look at the last two weeks of data or a week from each of the last three months depending on your seasonality issues.
    Take a look at the AHT times on a control chart of about 100 – 200 calls and see if your are in stability or if you have some special cause issues.
    If there is no special cause, or you can remove the special cause, I would then take a look at the distribution of my data to better understand the best tools to take a look at your current state, normal, non-normal, etc….
    Once you have a baseline DPM / Sigma Level, you can determine where you need to focus the attention and if it is truly worth while to continue based on your service level agreements with your customer or internal specs.
    Recapping thus far, sample about 200 calls randomly for a time that makes sense to you based on your type of call center.  Look at a control chart to check for special cause or non-random types of behavior.  Once special cause is removed or determined common cause, take a look at your data distribution and see what you got.
    I then like to stratify call types to see which ones have the largest variance or mean levels to know where to target my approach.  SO if all of your calls are not the same, like customer service and sales, then you need to see where you should focus your efforts, box plots and ANOVA studies are useful for determining differences between calls.  Make sure you have normal data or you must use non-parametric type analysis tools, or resample you data because of issues with the sampling method (most types of calls are normal if you do it properly from my experience)
    Once you determine what type of call to take a look at, I usally do 2 types of studies.  I look at the switch data output and will sample at random around 30 -100 calls to form my basis for inferential type tools.  Sampling techniques are important here, so make sure you read up on the best type for you.
    I then take a stop watch down to the line and check at random calls to verify that the switch is outputing properly and I see the same type of results based on my own sampling, (gauge study in a sense)
    Once you have your population determined and a good data set.  Take a look at what part of the calls show the most variablity or highest ammount of time spent.  If your newly sampled calls seem in control from a CC, then you will need to change somthing in your process to make the call types go down.
    ex. break down the AHT into is subcomponents and see what takes the most time or has the most variability. (example AHT = ACD+OBC+IBC+AUX+FollowUp+ACW)
    That is how I do it, good luck…
    Joe

    0
    #138795

    Whitehurst
    Participant

    I disagree with the University of Michigan judgement, Pat Hammett is an excellent Black Belt /Master Black Belt.  Between my last two six Sigma jobs I have had no problem keeping up and exceeding the instruction from the Michigan Program.
    In addition the Lean program at Univeristy of Michigan is taught by Jeffery Liker, Toyota Way Author, who I think has the ability to say he is an expert in the field. 
    Michigan has the second Highest rated program in the Country.
    But I will say, the flaw lies in the fact that anyone can take the program, so you will see a large range of abilities from the course.  I have experienced several U of M BB that are not to skill level.  This I attribute to the person not the program.  I would be interested in seeing any program that only produces the best Black Belts everytime, it is the person, the training only gets you half way.
    I have had both U of M and Ford training, full blown programs, I think the U of M one is far superior.  But there is always more to learn.
    Do what works for you, stay away from the programs that are online only, they are a problem in most cases, we call the Sigma Lites at my company.
    just my thoughts
    Joe

    0
    #64346

    Whitehurst
    Participant

    I am removing any copyright and proprietary info, I will post soon
     
    Joe

    0
    #138501

    Whitehurst
    Participant

    Call Center,
    I am a call center Black Belt and have tackled the idea that you are currently experiencing.
    I do not like sigma levels for our type of attributes because we often have more defects then a good sigma level will allow or show.
    I prefer DPM and DPMO.
    example
    total calls out of SL / Total calls = DPM * 1000000
    but if you must
    process Sigma = NORMSINV(1-((Total Defects) / (Total Opportunities))) + 1.5
    I also believe that it means more to people with no clue what a Sigma level is, I run across that alot in the Call Center enviornment.

    0
    #138492

    Whitehurst
    Participant

    Interesting question,
    You right about most companies, they value the certification over the knowledge up-front.  I recently accpeted a new job and was suprised by the lack of questions that I was asked in the hiring process.  I was prepared for a theoretical grilling of sorts, never came.
    They assumed since I was certified, that I must know the material.
    The person who hired me, my boss, is a Certified Black Belt as well, however, her certification was in the form of a 1,000 fly-by-night program that really lacked in depth knowledge.
    I myself, was trained by a Master Black Belt at Ford, but actually got my certification from a local University.  Their program was amazingly in-depth, the instructor was passionate about his teaching, and the content was sound.  I have ran across better black belts then myself of course, but not because they had better training, the reason being their raw experience.
    If I had not had a manufacturing job that I could apply my daily lectures to, I am not sure if it would have been as good.  Every lecture I learned how to do somthing I already did, much better and more process oriented.
    My boss however is what I like to call “Sigma Lite”, practically of no use to me or my colleauges in any way.  We somtimes will test his skill with some questions that we know, just to gauge his training, YTD he hasnt proven worthwhile.  I recently hired a Green Belt to assist me in my day to day data collection and analysis, I grilled him with both theory and conecptual questions, in addition to some sound statistical ones.  A real black belt, in my mind, is continuosly improving himself through study, practice, training, and reading.  I like to read 1 book a week if possible on things that I may not know or want to better understand.
    In closing, when comparing Black Belt to Black-Belt, make sure to really look close, as they are not all created equal.  Some, like myself, enjoy the data and process with a passion to get better, money is also nice but not the driving factor for me.  I latch on to other black belts that I know are better then me so I can learn there tricks and unique abilites to analyze data from different perspectives, everyday I learn somthing new.
    Certification is good, it will help you get a job, but to truly succeed and excel at that position, the Certification is nothing more then a piece of paper, you must use it to your advantage.
    I myself am a big fan of lean manufacturing, I am not certified, but know more the most due to my dilligence in studying and applying it whenever possible.
    Joe

    0
    #138358

    Whitehurst
    Participant

    well I am talking about service industries specifically, the only one I have heard of is HSBC that actively uses the methodolgy here, other multinationals do more of best practise implementation and I am not sure if any of the local organizations have actively deployed it…and the post was not intended to point fingers rather understand the local service industry in the region and learn how they look at quality practices/methodologies

    0
    #138276

    Whitehurst
    Participant

    Thanks for the inspiration guys.   It was nice to hear others have the same struggle I do as well.  I have had the take it all, or dont bother me approach, I should probably be more flexible.  I am just afraid that if I back off of my methodology, then they will turn me into there personal statistics slut like they did past Black Belts. 
     Things are better today again.  Still I get more push back then I ever imagined was possible.  I am rolling out a infrastructure very soon, as MBB myself, and my current Green Belt is in the U of M BB training as well.
    On friday i got like 9 requests for green belt training out of the blue….
    So anyway, the main director I have a problem with I think may be close to the end of days with all the slop I am providing in projects from his responsibilities.
    and to boot the SOB is an industrial engineering and still does not get why I do what I do.  Demming would bitch smack this character.
    I have got a few projects completed from AHT analysis, Attrition Analysis, Forecasting eWorkforce capability,
    I would be willing to share reports with people if they have some reports to share back.  Let me know if you are interested.
    Joe

    0
    #138114

    Whitehurst
    Participant

    Mike,
    Do I run a paired t Test? or do I establish a Hypothesis that the scanned test is more accurate and run a p Test?
    Regards

    0
    #138112

    Whitehurst
    Participant

    I like that reply WW, it is well put and to the point.

    0
    #138104

    Whitehurst
    Participant

    Mike,
    I have the real count and I know when they counted how many errors they had with the manual count and how many errors they had with the scanned count. (an error is when the count did not matcht the actual).
    Knowing the number of errors on the manual count and the number of errors on the scanned count, do I just take the average of the errors accross many samples and assume that the lower average is more accurate?

    0
    #138096

    Whitehurst
    Participant

    Thank you Mike for the reply.
    I am trying to justify droping one method of taking inventory but looking at the historical data and identify which has been more prone to errors.  If I count the number of errors in the manual count and the number of errors in the scanned count, can I compare the two distribution of the two error counts and determine which has been more accurate in the past?
    I hope I am making sense here.  I apologize if not.
    Regards,
    Joe

    0
    #64330

    Whitehurst
    Participant

    A hypothesis test will only tell you how your populations look compared to each other.  Equal, less, greater.
    You can use them to determine what kind of population fit you have, normal, weibull, etc..
    I am not really sure what good comparing the means and variance of (F and T test) of AHT and Hold.
    You can really only say, AHT is (,=) Hold Times using hypothesis tests.  This will not give you a good explanation of how one effects the other, just a method of looking at their populations and describing them compared to each other.
    Hope this helps,
     
    Joe

    0
    #64328

    Whitehurst
    Participant

    Arun,
    I am a call center SS Black Belt as well.  I do what you are speaking of quite often, but am not really using hypothesis test in the strict sense.
    Once you have a good ided of your popoulation distributions, (normality assumption) you should also pop the data into a control chart and look for special cause points that our out of spec with the rest of the group.
    If your data is normal (or assumed normal) and you have a process that is in control statistically from the control charts.  (I usually set up a random sample too for the individual call measurments rather then actual averages) you can use a regression analysis (fitted line in mintab) to take a look at the effects each other have.  The idea is for a regression, if you have an Rsq value of around (70% for me rule of thumb) you can state that the ammount of variation in A can be explained for 70% by the variation in B (AHT vs Hold Time)
    Once you have established that there is a correlation, you need to dig down and take a look at the residuals:
    Residuals vs. Fitted, Distribution of Residuals, look at the best fit for normality and make sure you see a good fit for the residuals.  In minitab there is a button for 4 in 1 graphs that will contain all of this when you run the regression.
    But now let us look at it from a overview standpoint.  Hold time is a component of AHT, Right.  The basic formula for AHT takes into account Hold Time.  Thus you would have some automatically built in correlation for a certain ammount.
    I usually break down the AHT in a Excel Bar Graph that shows all of the different proportions in one bar graph. AHT = ACD time +outbound call time + After Call Work(ACW + etc) it is different based on where you work.
    So looking at AHT as the denomintaor, everything else in the numerator, what proportion of AHT is taken by each call, this will give you a stronder understanding of the system I think then a regression at this point.  If you do choose to run a regression, grab a stats book to make sure you assume all the proper requirments to determine if the regression is indeed significant
    hope this helps.

    0
    #138040

    Whitehurst
    Participant

    I as well like the 5/3/3 it has proven most reliable in manufacturing measurments.  Take a look at the minitab function if you have it available, or if you would like, i can send you my home brew excel one we used at Ford.

    0
    #138037

    Whitehurst
    Participant

    Thanks for the posts.
    I have only worked for two companies in the past 5 years.  I agree with the job hopping problem.  I do not wish to leave my current position as it is an amazing challange full of potential for my portfolio.  I just had a real bad day yesterday and needed to vent it out.
    Im looking for more of a approach that people will accept the concepts and not make decisions on “I think” type mentalities.  Regardless of my frustration, I have changed the company dramatically from an operations side, it is more getting to that point that I find difficult.  When I was employed at Ford, there was a wonderful structure in place and the culture was accepted and highly sought after, there were challanges, but the support from above was amazing.
    I will have saved the company 2 million by the end of the year if my changes hold pace and are controlled as identified and signed off upon by our CEO.  We are only a 90 million dollar company, with 10 million total profits, so 2 million is a 20% increase in our bottome line.
    I saw this as an opportunity to deploy Six Sigma in a company that needs it badly, and this is what I was promised upon  hiring.  They will pay for my MBB training since it is my contract, that is enough for me alone to stay along and fight.
    It seems my clients are the only one who appreciate me at the end of the day.  Which I guess I should cling to since they are the ones who really pay the check.  My most recent analysis uncovered alot of falacies in the company vs the statement of work (SOW) and that alone I think has proven my worth.
    Nonetheless, it is still a major challange that I must face.  The “speak when you are spoken to” comment just set me over the edge yesterday.  I think that alone was the reason I posted with such fervor.
    I am working from a deployment standpoint and maybe I need to start getting some executive overview and champion trainings in place for future ease of project adversity.  I am also piloting a white belt project to get us a ground up approach.  Because of our high turnover, a full blow greenbelt is not feasible, but if they work up slowly from white, maybe that will help, time will tell
    Thanks for the insights and if anyone has some good reccomendations on how to get some more positive visibility in the company from personal experience, I would love the comments. 
    Joe

    0
    #135981

    Whitehurst
    Participant

    Thanks for that thought. I was pulling a rookie mistake on the distribution identification.
    What size sample / time in your opinion(expereince) would be good to look at to draw a group wide conclusion?
    Like sample 30 AHT in the day and use it, I have averages available to me very readily but am scared as they hide the true variation.
    Good thoughts though.  I agree on the Imr, the chart for the Xbar is nasty due to subgrouping constraints.  I would like to think a little further though.  My ranges are really nasty between operators with constant calls and this obviously biases my ouput in an IMR char because three sigma up or down is alot when you are talking about a sigma of like 100 or 300 seconds.
    Yeah, I like my widgets.
    Joe

    0
    #135968

    Whitehurst
    Participant

    I agree with the Y being # of calls.  The process is so far out of control that I have never seen this type of variation in anything.  I agree also with the project to take a look at variables such as call length, after call work.
    The only real measurment target we are tasked with is Service Level.  Service level is the % of call answered within a certain time. (ex 70% of calls within in 30 seconds)
    Mabye a more representative task for Y is # of calls being answered within service level.  In addition I can make a proportion defect style control chart where calls answered below a certain service level are automatically binned as bad news. (ok I like that Idea, amazing how a simple post on Isixsigma is answering my own questions logically)
    From an 30k level metric, the process is lieing to me as averges are very nasty in this world.  I obviously need to start rocking some interval data for more specific approaches to my world.
    There is also a tremendous gaging problem when it comes to the issue types an operator can select for a disticnt call.
    But back to control charts.  Is an IMR acceptable to monitor Average Handle Time per day/week/month for an operator. (no subgroups for the daily stuff
    Joe

    0
    #133315

    Whitehurst
    Participant

    Could you e-mail me the info?

    0
    #132623

    Whitehurst
    Participant

    The answer is :
    Everything has a variation even six sigma  now whether its 1.5 as a constant thats the debate.
    joe
     

    0
    #132553

    Whitehurst
    Participant

    Hi PGK,
    A problem, in your case customer claims, brought about by human factor is something that is not very easy to handle and indeed, promotion of mistake- proofing or poka-yoke devices is a very good way to tackle this. Of course, not all operation systems could be made automatic, installing devices, sensors, etc. as these are costly but there is Visual Control and 5S. Going down to the human aspect , one must check too the motivation factor of the person involve. Sometimes, when all the “why’s” are all used up, it boils down to the aspect of behavioral training. Even though how good a procedure is, it is still at up to the operator if he would do it or not.
    cheers,
    joe

    0
    #132199

    Whitehurst
    Participant

    C Chart is used for the number of defects per unit. The number of defects are plotted directly in the control chart .
    Cheers!
    joe
     

    0
    #131453

    Whitehurst
    Participant

    the values seems confirm what you said and the slip you obeserve could be due to the fact that data sets come from similar population, but from different side of distribution. If could explane better which is the scope of your test and raw data, maybe we can understand which are some further test you can run to confirm your hypothesis, because having data set (samples , I think) of equal size, it is possible  to run also other test that could help.
    regards, joe

    0
    #131395

    Whitehurst
    Participant

    Perform an analysis of variance and try to understand from which side the variation come from and then start to think about which action to put in place to correct it.
    Regards, joe

    0
    #131394

    Whitehurst
    Participant

    Could you say qty for each set of data and chisquare value from mood’s median test ?
    Regards, joe

    0
    #131337

    Whitehurst
    Participant

    CTQ – > Average Handle Time

    0
    #130870

    Whitehurst
    Participant

    Yes, my mistake. I worte wrong. Your value is right. It is 1 tail T value with alpha=0,05 and df=9.
    Regards, joe

    0
    #130855

    Whitehurst
    Participant

    Filex, in excel if you want T value for 2 tails with alpha=0,05 and df=9 you must write Tinv(0,05;9)=2,262159. If you want T value for one tail with alpha=0,05 and df=9 you must write Tinv(2*0,05;9)=2,262159
    The same for alpha=0,025.
    2,68501 is T value for 2 tails with alpha=0,025 and df=9 wrote Tinv(0,025;9).
    Hope this help
    Joe

    0
    #130790

    Whitehurst
    Participant

    Maybe we are doing confusion on terms.
    It is well know, from reference tables also,  that with degree of freedom=9,  the T values for alpha=0.05 for 2 tails  is the same of  T value for alpha=0.025 for 1 tail and viceversa.
    I hope this is the answer that explane your concern, anyway to better understand your calculation concern with excel and minitab, you could provide the details of settings.
    joe

    0
    #130736

    Whitehurst
    Participant

    You have to specify the tails and test type, both in excel tan minitab.
    regards, joe 

    0
    #129716

    Whitehurst
    Participant

    Just a suggestion : why do you want to use x bar chart. why dont u use individual chart instead if the average as you say is misleading.

    0
    #129715

    Whitehurst
    Participant

    according to Chebycheb’s theorem for a noraml shaped distribution 99.93% of the data fall within +- 3 standard deviation so any point lying outside the control limits has a probabilty of only.027 ( which we term as a special cause ).Thats the reason we use 3 not 4 or 5 or 6.
    We use p chart in case in case of number of defectives and when the subgorup sizes are not equal.So it is used to determine if the rate of non conforming product is stable and detect when the deviatiion from the stability has occured.
    UCL / LCL = p bar + 3sq root ( pbar qbar /n)
    Hope this helps!!
    Regards
    joe
     

    0
    #129040

    Whitehurst
    Participant

    CT you are right. This is the point. A poor management planning (how design a factory), become a people problem,  too time waste at bathroom. I think your post is the real whole point. Could be interesting to calculate $ spent along years for that and then the project could be to cut management salary, not time for bathroom.
    joe

    0
Viewing 100 posts - 1 through 100 (of 285 total)