THURSDAY, OCTOBER 30, 2014
Font Size
Members MBBinWI Activity

MBBinWI

  • As my buddy @cseider says, don’t worry about the shift. 1) It’s not 1.5 (however, it does exist but it is unlikely to be 1.5 – it will be unique for each measured item), and 2) If you are monitoring your process and adjusting (not over-adjusting, but properly adjusting) then you can reduce that overall shift dramatically.

    How did you select 1.33…[Read more]

  • Matt: At the GB level, you shouldn’t have any issues with statistics. As long as you understand some of the basics – mean, median, mode, what a standard deviation is and why it is important, and the basics about distributions. As @rbutler identifies, knowing regression and control charts will be a great aid. For GB’s, graphical analysis should…[Read more]

  • @Cor1 – some good advice you’ve received here. What I haven’t seen discussed is the need for a “burning bridge.” Do you have some aspect of the business that is struggling mightily? For a production or service company, is there a product or service or facility that the organization is contemplating terminating? Or, alternatively, one that is…[Read more]

  • @Helmet – Instead of asking for a “problem” you might approach the process owners with the query – “What takes too much time in you process to …?” or alternatively, “Where do you experience excessive rework (or redo)?,” or even “What is preventing you from doing more/closing faster?”

    As my friend @cseider states, it’s too late now, but a…[Read more]

  • MBBinWI replied to the topic Hypothesis Question in the forum Methodology 2 months ago

    @cprao101 – just as a friendly suggestion, in such a situation, it would be better for you to posit an answer and your reasoning. That will likely get you a positive response.

    You were lucky that Katie got to you first. There are many more of us who wouldn’t have likely been so gentle ;-)

  • GB: Long term must be no better than equal to, and will normally be less than the short term. The real question comes as to what constitutes “short” and “long.” Regardless of how you define them, since “long” encompasses the possibility of things occurring that would not occur under “short” there will be more variation and thus a lower Cpk.

  • Jose: You’ve received some sage advice here. You are correct that a single value for Cpk isn’t always the best. However, it is often necessary to drive an objective with a simple standard. You should look at the costs of improving and the benefits of improving and make a reasoned argument as to the return on that investment. Good luck.

  • As my buddy @rbutler identifies, you need to look into the situation of the material effects. This looks like a split plot type design.

  • @itsmi – The reason that my friend @cseider suggests doing some analysis on the effects of the X’s (the independent parameters that you set or possibly affect the output) on the Y(‘s) is two fold. 1) To see if they really do have some effect. 2) To see at what level of change they have a significant effect. For example, you mention humidity. …[Read more]

  • @cseider – I’m shocked! ;-)

  • @venu_creative – To the excellent response by @rbutler I would add that if you can establish that Y=f(x’s) equation, you can use partial differentials to identify the influence of the x’s in comparison to the others.

  • @cseider – CrystalBall is a much better MonteCarlo simulator.

  • @Johnnynita – for crying out loud, this is a classic training problem. Do you really want us to answer your homework problem?

    I’m surprised my friend @cseider even offers an opinion other than to chastise you to do your own homework.

  • @liamocol – are you using enough decimal places?

  • @7sigma – I would expect that this has more to do with the culture of the organization. Those that are used to change will find it less of a problem than those that change infrequently.

  • @wchepul – You also will need to evaluate the level of non-conformance and the risk you will accept in not being able to estimate “closely enough” to the actual level.

    Look at it this way, if you have 50% defective, it won’t take very many samples to identify an estimate on the overall population defect rate (within some acceptable predictive…[Read more]

  • Hey, Z: Not sure if you are still looking for some help on this, but I’d check into using a Chi square.

  • MBBinWI replied to the topic OEE / OPE in the forum Manufacturing 4 months, 1 week ago

    Kenneth: Look at the components of OEE and do a Pareto analysis on the machines and fault codes/reasons for being less than 100%. That will at least give you a target list for improvement opportunities. At that point, you can use a cost/benefit type analysis to determine whether to attack that issue or not.

  • @Mike-Carnell – I suppose offering my house as a “learning lab” would be too selfish?

  • MBBinWI replied to the topic OEE / OPE in the forum Manufacturing 4 months, 1 week ago

    @Mike-Carnell – There you go, trying to make sense. ;-)

  • Load More

Login Form