iSixSigma

Taguch Vs Clasical

Six Sigma – iSixSigma Forums Old Forums General Taguch Vs Clasical

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
    Posts
  • #29509

    PA
    Participant

    Can someone help me out to understand the difference between TAguchi approach VS Clasical fractional design .

    0
    #75801

    Mike Carnell
    Participant

    PA,
    Two things are going to happen here. First you are going to get a lot of detailed statistical answers. There are large factions who are Taguchi bigots – the only way to do it. There is a large faction that hates Taguchi designs – equally as ignorant. You can’t ignore the success. AS a tool it has to be taken seriously.
    The second thing is I will give you my version of the difference and I will get hammered as well. Here goes – basically Taguchi is an extremely fractionated design.
    The best book I have seen on the mixture of different types of designs is “Straight Talk on Designing Experiments” by Bob Launsby and Daniel Weese, ISBN 0-9636093-3-5. It is subtitled “Blending Taguchi & Classical Approaches.”
    Good luck.

    0
    #75807

    Monte Bainter
    Participant

    I believe Genechi Taguchi himself would tell you that all he has done was to apply Fractional Factorial DOE as a practical tool for industry. 
    For me, I look at the intent of the undertaking.  If I am looking to report conclusions about a number of the tested factors (when we do X, then THIS happens and when we do Y, THAT happens – but if you really want to see something, take a look at what happens when I do THIS to W while standing on my left foot and whistling Dixie), I head down the classical path. 
    If I am looking for the very clear, singular, stated purpose of Taguchi – What equipment settings and process factors will provide optimum output? – I default to Taguchi.  Taguchi developed a concise, understandable, trainable method (including design arrays) to serve that single purpose.  Was it Kume (or perhaps even Taguchi) who expressed disbelief in America’s fascination with (and time/effort spent) trying to understand the details of how and why things DID NOT work – when all we really want to know is simply “How should I run this”?  Taguchi simply addressed that irony by preparing and presenting a practical tool for industry. 
    An important note to consider regarding Taguchi (or fractional classical for that matter) that many don’t ever quite grasp is this:  Your experiment indicates Pressure (e.g.) to be a 70% contributor to successful results.  So you tend to then walk around telling people that “It’s all about Pressure.”  But that significance is a direct result of the range between limits you selected.  If the difference between limits had been only a fraction of that of the original experiment, Pressure might not have even shown up as a contributor at all and now Temperature is a 50% contributor.  “Significant” within the tested range only.  (I know you didn’t ask for that last part, but I have to throw it in anytime Taguchi comes up.)

    0
    #75808

    Kim Niles
    Participant

    Dear PA:
    Addressing your question literally, most Taguchi designs are the same as “classical fractional designs” and even full factor designs such as the L8 = 2^3.
    Reading between the lines, Taguchi took simple designs that existed and added a few other simple designs that utilize easy to make assumptions of non-existent higher order interactions to compile a group of matrices most experimenters would typically use.  I like to use them when I already know a lot about the process but I still need to screen out special cause variables. 
    Two potential problems exist with them.  One is that you are making more assumptions and so are trading resources for risk.  The other is that they are usually much lower resolution designs and so you need to follow up with confirmation experiments or lower your alpha risk requirements (i.e. you are not as likely to get significant results with very high confidence but sometimes that doesn’t matter).  
    I hope that helps.
    Sincerely,
    KN https://www.isixsigma.com/library/bio/kniles.asp
     
     

    0
    #75809

    PA
    Participant

    thanks  Mike , Kim , Monte . your messages have been very useful .
    I will also see the suggested book
     

    0
Viewing 5 posts - 1 through 5 (of 5 total)

The forum ‘General’ is closed to new topics and replies.