Re: [MBZ] Bathtub and bell curves
Edward, The feedback was general & directed to everyone. The bell curve is more correctly called a standard, normal curve. There is also a proper name for a bathtub curve but I don't remember it right now. The right side of the curve may take on the shape of a bell curve for items like a light bulb, in this case it's a standard normal curve & the early failures are termed outliers. But with some products, the line continues to rise when MTBF is approached & with others, the line flattens out when MTBF is approached. In these cases, the curve can-not be made to resemble a standard normal curve, regardless of the scale used. Thanks, Tom Hargrave www.kegkits.com 256-656-1924 -Original Message- From: Robert Bigham [mailto:[EMAIL PROTECTED] Sent: Sunday, August 05, 2007 6:50 AM To: Tom Hargrave Subject: RE: [MBZ] Bathtub and bell curves Hello Tom "By the way, a bell curve & bathtub curve do not complement each other. A bathtub curve tends to be more one sided, with most failures on the tail end. A bell curve is balanced." - Tom If you say so. I have been wrong or mistaken before; it is/will not be a new experience. But then, I'm not a statistician, I'm just a guy who uses simple statistics in engineering applications. Statistics is the science that deals with the art of making decisions in the face of uncertainty. I usually merely need to make sense of data which approximate a normally distributed random variable, so the high powered statistical methods are not a regular part of my thing. I presume you shared your insight on this with Mitch also, who told us about the bathtub curve, thus: " More like the antithesis of the bell curve. A 'normal distribution' bell curve peaks at the mean and drops off on both sides. The 'bathtub' as he describes it, has peaks on both ends and a wide trough in the middle." - Mitch. There was someone speaking to this before Mitch, I know not whom. The one sided curve you describe sounds to me something like a sort of chi-squared distribution with a good many degrees of freedom. Or perhaps it is no more than a comparison of time to failure versus cumulative number of failures. All things fail in the end. Nothing lasts forever. Some things come closer than others. I don't find anything called "bell" or "bathtub" in my old statistics references. Presumably these are laymen's terms describing the density function of the normal distribution in the bell case and something else, I really know not what, in the bathtub case. As defined by Mitch, the bathtub case sounds like a simple upside down plot of a normally distributed random variable, which we must all admit would look like what Mitch described. We may always plot and display our data however we wish. In my old academic discipline, some types of data are conventionally plotted always in the second quadrant. I know of no particular reason why that is a better presentation or leads to more insightful analysis. It's just done that way. But it is all fun. Have nice day. Robert . > [Original Message] > From: Tom Hargrave <[EMAIL PROTECTED]> > To: <[EMAIL PROTECTED]>; Mercedes Discussion List > Date: 8/5/2007 12:45:01 AM > Subject: RE: [MBZ] Bathtub and bell curves > > Robert, > > "Clear as mud". > > Understanding MTBF is the reason I searched for a 70's Corvette with about > 50,000 miles on the clock when I decided to buy one. Even though my Vette is > a 1977, I know that all of the major modules and assemblies are in decent > shape. The rubber parts are another story. > > Mercedes on the other hand are much better designed & I have no problem > purchasing one with 200,000 miles on the clock, providing everything (and I > mean everything) works when I buy the car. I understand that some assemblies > are approaching end of life but I also know that quite a few have been > replaced & quite a few will go another 100,000 miles or so. I also don't > have a problem with putting a Mercedes with 200,000 miles on the clock on > the road, but only after I've gone over all of the systems. > > Compare the 1977 Corvette to the 3 - 5 year water heater & a same year > Mercedes to the 12 year water heater discussed earlier. > > By the way, a bell curve & bathtub curve do not complement each other. A > bathtub curve tends to be more one sided, with most failures on the tail > end. A bell curve is balanced. > > Thanks, > Tom Hargrave > www.kegkits.com > 256-656-1924 > > -Original Message- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] > On Behalf Of Robert Bigham > Sent: Saturday, August 04, 2007 11:02 AM > To: mercedes@okiebenz.com > Subject: [MBZ] Bathtub and bell curves >
Re: [MBZ] Bathtub and bell curves
Robert, "Clear as mud". Understanding MTBF is the reason I searched for a 70's Corvette with about 50,000 miles on the clock when I decided to buy one. Even though my Vette is a 1977, I know that all of the major modules and assemblies are in decent shape. The rubber parts are another story. Mercedes on the other hand are much better designed & I have no problem purchasing one with 200,000 miles on the clock, providing everything (and I mean everything) works when I buy the car. I understand that some assemblies are approaching end of life but I also know that quite a few have been replaced & quite a few will go another 100,000 miles or so. I also don't have a problem with putting a Mercedes with 200,000 miles on the clock on the road, but only after I've gone over all of the systems. Compare the 1977 Corvette to the 3 - 5 year water heater & a same year Mercedes to the 12 year water heater discussed earlier. By the way, a bell curve & bathtub curve do not complement each other. A bathtub curve tends to be more one sided, with most failures on the tail end. A bell curve is balanced. Thanks, Tom Hargrave www.kegkits.com 256-656-1924 -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Robert Bigham Sent: Saturday, August 04, 2007 11:02 AM To: mercedes@okiebenz.com Subject: [MBZ] Bathtub and bell curves The question of how long a light bulb lasts is used in several introductory books on statistics. How long does a light bulb last? A truthful answer is "It varies." But people tend to want a single number. Maybe the average. As applied to this problem, the bell curve plots bulb life (in hours or some unit of time) against number of bulbs in the sample that last so long before failing. The bathtub curve plots bulb life against number of bulbs failing at some particular life. The curves are something of complements of each other; at any life the two add up to 100% of the bulbs in the experiment. Some fail out of the box.or early, then few fail, then at the end, all survivors finally fail. Conversely, only a few last a short time, many last something like the average life, and only a few last into light bulb old age. Clear as mud? Robert > Message: 6 > Date: Sat, 04 Aug 2007 07:17:16 -0400 > From: Mitch Haley <[EMAIL PROTECTED]> > Subject: Re: [MBZ] Tools and Quality > To: Mercedes Discussion List > Message-ID: <[EMAIL PROTECTED]> > Content-Type: text/plain; charset=us-ascii > > > Zoltan Finks wrote: > > > > Man, that gives me flashbacks of the statistics class I took to get my > > bachelor's degree recently. Is the bathtub curve the bell curve? > > More like the antithesis of the bell curve. A 'normal distribution' bell > curve peaks at the mean and drops off on both sides. The 'bathtub' as he > describes it, has peaks on both ends and a wide trough in the middle. > Mitch. ___ http://www.okiebenz.com For new parts see official list sponsor: http://www.buymbparts.com/ For used parts email [EMAIL PROTECTED] To Unsubscribe or change delivery options go to: http://okiebenz.com/mailman/listinfo/mercedes_okiebenz.com ___ http://www.okiebenz.com For new parts see official list sponsor: http://www.buymbparts.com/ For used parts email [EMAIL PROTECTED] To Unsubscribe or change delivery options go to: http://okiebenz.com/mailman/listinfo/mercedes_okiebenz.com
[MBZ] Bathtub and bell curves
The question of how long a light bulb lasts is used in several introductory books on statistics. How long does a light bulb last? A truthful answer is "It varies." But people tend to want a single number. Maybe the average. As applied to this problem, the bell curve plots bulb life (in hours or some unit of time) against number of bulbs in the sample that last so long before failing. The bathtub curve plots bulb life against number of bulbs failing at some particular life. The curves are something of complements of each other; at any life the two add up to 100% of the bulbs in the experiment. Some fail out of the box.or early, then few fail, then at the end, all survivors finally fail. Conversely, only a few last a short time, many last something like the average life, and only a few last into light bulb old age. Clear as mud? Robert > Message: 6 > Date: Sat, 04 Aug 2007 07:17:16 -0400 > From: Mitch Haley <[EMAIL PROTECTED]> > Subject: Re: [MBZ] Tools and Quality > To: Mercedes Discussion List > Message-ID: <[EMAIL PROTECTED]> > Content-Type: text/plain; charset=us-ascii > > > Zoltan Finks wrote: > > > > Man, that gives me flashbacks of the statistics class I took to get my > > bachelor's degree recently. Is the bathtub curve the bell curve? > > More like the antithesis of the bell curve. A 'normal distribution' bell > curve peaks at the mean and drops off on both sides. The 'bathtub' as he > describes it, has peaks on both ends and a wide trough in the middle. > Mitch. ___ http://www.okiebenz.com For new parts see official list sponsor: http://www.buymbparts.com/ For used parts email [EMAIL PROTECTED] To Unsubscribe or change delivery options go to: http://okiebenz.com/mailman/listinfo/mercedes_okiebenz.com