High quality statistics - diyAudio
Go Back   Home > Forums > Member Areas > The Lounge

The Lounge A place to talk about almost anything but politics and religion.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 28th October 2014, 02:32 PM   #1
diyAudio Member
 
auplater's Avatar
 
Join Date: Apr 2005
Location: KyOhWVa tristate
Default High quality statistics

Some "statistical analysis" to ponder next time you look for causation:

Spurious Correlations

Click the image to open in full size.

Click the image to open in full size.

etc., etc.

John L.
__________________
"...His brain is squirming like a toad..." Jim Morrison
  Reply With Quote
Old 28th October 2014, 04:59 PM   #2
diyAudio Member
 
Join Date: Jan 2013
Location: Oregon City, Oregon
Many years ago, I had a subscription to a publication, "The Journal of Irreproducable Results". My favorite article was about "Data Enrichment", where they used statistics slight-of-hand to get an excellent correlation between height above floor and "heads" on a coin toss...
  Reply With Quote
Old 28th October 2014, 06:17 PM   #3
SY is offline SY  United States
diyAudio Moderator
 
SY's Avatar
 
Join Date: Oct 2002
Location: Chicagoland
Blog Entries: 1
You've ruined my productivity this morning, John.
__________________
And while they may not be as strong as apes, don't lock eyes with 'em, don't do it. Puts 'em on edge. They might go into berzerker mode; come at you like a whirling dervish, all fists and elbows.
  Reply With Quote
Old 31st October 2014, 03:43 PM   #4
Glowbug is offline Glowbug  United States
diyAudio Member
 
Glowbug's Avatar
 
Join Date: Feb 2006
Location: Hillsborough, NC
Click the image to open in full size.

Clearly...
__________________
Jim
The machine does not isolate us from the great problems of nature but plunges us more deeply into them. - Antoine de Saint-Exupéry
  Reply With Quote
Old 31st October 2014, 03:54 PM   #5
rdf is offline rdf  Canada
diyAudio Member
 
rdf's Avatar
 
Join Date: Jun 2004
Location: big smoke
I say we play it safe. For the children.
Attached Images
File Type: png q54sO25.png (35.9 KB, 125 views)
__________________
Ears aren't microphones.
  Reply With Quote
Old 31st October 2014, 05:33 PM   #6
diyAudio Member
 
Join Date: Nov 2009
Location: Birmingham, UK
Let's just assume causation for a moment:
Do people drown in pools because Nicholas Cage appears in movies or does Nick get roles because enough people drowned?
  Reply With Quote
Old 31st October 2014, 06:28 PM   #7
diyAudio Member
 
jacco vermeulen's Avatar
 
Join Date: Oct 2004
Location: At the sea front, Rotterdam or Curaçao
Send a message via Yahoo to jacco vermeulen
Let me put it like this : Eli Lilly decided it was more productive Not to renew the Prozac sponsorship deal with Mr Cage.
Without the additional income, he was forced to accept more roles.
__________________
The buck stops Here
  Reply With Quote
Old 2nd November 2014, 01:14 PM   #8
diyAudio Member
 
tubelab.com's Avatar
 
Join Date: Jul 2005
Location: West Virginia panhandle
In the 1970's Motorola's product quality was pretty bad, as was the quality of pretty much every American company. The Quasar TV's were so bad that they sold the entire division to Panasonic and got out of the business. Freshly built products rolling off the assembly line didn't work, so technicians like me were hired to fix them. The Defects Per Unit (DPU) rate was a number greater than 1. This was expensive.

This started a revolution (1980's) in product design and manufacturing. Within 10 years the DPU rate had to be abandoned for the DPHU rate (Defects Per Hundred Units) which was a number less than 1.

Part of this were mandatory classes for all engineers in Design For Manufacturing, Design For Assembly, Failure Mode Analysis, Statistics and More Statistics. All of this was part of a new "revolution" called Six Sigma Design.

We all learned to worship Sigma. Fancy analysis tools were created to appease Sigma. Sometimes entire programs were held up because Sigma wasn't happy.

We had a particularly tough product design involving a lot of new previously unseen and untried technology, in which many of the product design specs were arbitrarily set by a bunch of suits in a meeting room. Several prototype runs were done, after which the results were distilled down into "management quality" charts that only presented the CP and CPK graphs which were not meeting the required Six Sigma Design numbers. A whole bunch of yelling and screaming ensued, which begat meetings and more meetings to "empower teams" to go off and solve the problems. I got stuck on one of those teams where most of the members were not hardware engineers, nor fully understanding of 6 Sigma. We were assigned to solve some CPK failures that had been with the product since first proto run.

Too often the real issues were overlooked. CPK is a measure to see how well the distribution of your product's performance measurements matches up with it's specifications. Ideally all the measured data should lie centered within the spec window. CPK will not compute with a one sided spec like transmitter power must be greater than 6 watts, so an arbitrary upper spec limit is set, say 10 watts.

It turns out that these arbitrary upper spec limits were the problem with two of the big CPK failures, so I wrote an Excel Macro that I dubbed "The Six Sigma Shuffle" and ran it against these two specs. It just iterated the upper spec limit against the raw data and solved for the highest CPK. Bingo...CPK is now OK, but we have to fail a small number of radios because they are too GOOD!

We reported our results in the weekly management meeting, management is happy, and our crisis team is disbanded, and we go back to our regular jobs. About 2 days later an engineer comes to me and asks if I can make the Shuffle work on a two sided spec, so I modify it.

They run the shuffle against the entire collected data set for the last several prototype runs to determine the NEW product design specs in order to make Sigma happy. Many of the original product design goals were modified to fit the data on what we could actually build and the product took a giant step toward being one of the most successful radios to ever come out of the Florida factory. That was over 20 years ago and I still see those units in use today.
__________________
Too much power is almost enough! Turn it up till it explodes - then back up just a little.
  Reply With Quote
Old 3rd November 2014, 10:51 AM   #9
diyAudio Member
 
ruffrecords's Avatar
 
Join Date: Jun 2007
Location: Holt, Norfolk
Too often people mistake data for information. Many years ago I was called into a PMR manufacturer who was having a lot of end of line failures of their new transceiver. we sat on the line for a morning and watched them gathering data to pass to management, none of which pointed to the problem. Data from each stage of the line showed units passed but a significant proportion failed final test. In the afternoon we sat at each station of the production line and talked to the operatives. We very quickly found the problem. The rf front end included a complex helical filter which was aligned by technicians. We soon discovered some filters were very hard to calibrate according to the instructions provided by R&D but the technicians had learned how to' tweak' them so they passed. Thing is, the troublesome units were faulty (a single wrong capacitor value) which is why they failed the alignment procedure. The next day we sat with the techs and insisted they follow the procedure and failed units that would no align. They all turned out to be faulty.

Cheers

Ian
  Reply With Quote
Old 3rd November 2014, 12:53 PM   #10
diyAudio Member
 
jacco vermeulen's Avatar
 
Join Date: Oct 2004
Location: At the sea front, Rotterdam or Curaçao
Send a message via Yahoo to jacco vermeulen
Quote:
Originally Posted by ruffrecords View Post
Thing is, the troublesome units were faulty (a single wrong capacitor value)
QA/QC must have been foreign language in those days.
(textbook example)
__________________
The buck stops Here
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Output stage statistics akis Solid State 17 22nd June 2014 09:44 AM
USB socket comparison for high quality and high demand audio usage diyinhk Parts 2 12th June 2014 01:53 PM
Statistics Osvaldo de Banfield The Lounge 3 27th October 2012 10:13 PM
High quality, high voltage (~250V) reservoir caps for PSU thadman Power Supplies 5 10th May 2010 10:18 PM
Low wattage, high efficiency, high quality? Nordic Multi-Way 5 5th September 2005 12:12 AM


New To Site? Need Help?

All times are GMT. The time now is 01:07 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2