SMath Studio Forum
»
SMath Studio
»
Questions
»
Chain Failure Weibull Analysis
Rank: Advanced Member Groups: Registered
Joined: 29/11/2017(UTC) Posts: 82 Location: Midwest Was thanked: 9 time(s) in 7 post(s)
|
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 29/11/2017(UTC) Posts: 82 Location: Midwest Was thanked: 9 time(s) in 7 post(s)
|
The solve block is in the "Weibull Distribution" area (collapsed). nas
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: nasch I have tried to convert the attached MC15 worksheet to SMATH. Smath converts zero, absolute zero MCD.
You have to create the project from square one . At least, you can plot 3 graphs
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 29/11/2017(UTC) Posts: 82 Location: Midwest Was thanked: 9 time(s) in 7 post(s)
|
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 29/11/2017(UTC) Posts: 82 Location: Midwest Was thanked: 9 time(s) in 7 post(s)
|
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 29/11/2017(UTC) Posts: 82 Location: Midwest Was thanked: 9 time(s) in 7 post(s)
|
Chain Failure Weibull Analysis - 2.sm (53kb) downloaded 11 time(s).Taking jmg's advice; started going thru the steps and found a function that didn't come over from MC that I used the wrong "Sum" function in SMATH. Worksheet function ll.W(....) now seems to work plotting varying lambda and beta (sheet attached). I still cannot determine/maximize the function, finding the optimim lambda and beta. I welcome any suggestions. nas
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 28/08/2014(UTC) Posts: 1,356 Was thanked: 815 time(s) in 516 post(s)
|
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: nasch I still cannot determine/maximize the function, finding the optimim lambda and beta. Is that what you mean: that the n*F.W(x,λ,β) is the best fit to data {TY] ?Very rarely, a model function can be declared exact. That is the case of all LS based methods. I guess the data set [TY] can be fitted by many models, surely giving λ,β closely to Alvaro. Easy to plot [TY] over the n*F.W(x,λ,β) ... module added . Interesting but not of immediate understanding for the novice. Cheers ... Jean Opened in 6179, re-posted from same original. Chain Failure Weibull Analysis - 3.sm (72kb) downloaded 21 time(s).
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 15/04/2012(UTC) Posts: 1,986 Was thanked: 1124 time(s) in 721 post(s)
|
|
|
1 user thanked mkraska for this useful post.
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 29/11/2017(UTC) Posts: 82 Location: Midwest Was thanked: 9 time(s) in 7 post(s)
|
Alvaro, jmg and Martin, many thanks!
Martin's requires a newer version of SMATH installed, I'll have to request the upgrade to be installed.
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 15/04/2012(UTC) Posts: 1,986 Was thanked: 1124 time(s) in 721 post(s)
|
Originally Posted by: nasch Alvaro, jmg and Martin, many thanks!
Martin's requires a newer version of SMATH installed, I'll have to request the upgrade to be installed. Should work with SMath build 6671 as well (latest stable with updated plug-ins), just ignore the complaints. |
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: nasch Martin's requires a newer version of SMATH installed, Maybe NOT ... Maybe I'm too ignorant about Weibull. What I understand: over 'T', you have observed cumulative failure, is it ? From there you can approximate the probability of failures from the logistic Verhulst. I would be curious you find simpler and more realistic and more immediate insight. Done Smath 6971, should work in more recent SS. Cheers ... Jean Page0Weibull.sm (41kb) downloaded 20 time(s).
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: Jean Giraud What I understand: over 'T', you have observed cumulative failure, is it ? Not like said. The 'Y' data are the days between failure. Refreshed and added the cumulative integration of the Verhulst fit. Genfit Weibull.sm (46kb) downloaded 24 time(s).
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 29/11/2017(UTC) Posts: 82 Location: Midwest Was thanked: 9 time(s) in 7 post(s)
|
Thank you Jean, I have a lot to look over. I noticed that the MC file calculated a beta > than 1, which is a "wear-out" characteristic vs. the SMATH beta <1; "infant mortality". I found an earlier MC file that used a different method to find beta, and that too beta was < 1. Using excel, that was also in agreement; beta < 1. The values vary, the old MC file = 0.99, 0.98 for the SMATH file, and .99 for excel.
I use the files as a reference, being a novice using tools for stats. I did get a lot of new ideas and understand more using SMATH, which I am thankful to this collab.
Sincerely,
Norm
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 29/11/2017(UTC) Posts: 82 Location: Midwest Was thanked: 9 time(s) in 7 post(s)
|
Hello Jean, Beta will not calculate, I have the same SMATH version 0.99.6671. Maybe I'm missing a plugin?
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: nasch "infant mortality" ... because there 3 kinds of lies: 1. Kid's white lie [Mamy, Mamy ... the cat did suck your last candy] 2. True lies [Politician's lie] 3. Statistics What happens in the β "infant mortality" is consequence of reflexivity in the function wrt parameters. Many models can't be concluded by NO maths ... just manual from St. Patience ! If you stick with Weibull failure analysis, good luck. In about ½ hr I found nothing from Gooling.
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 29/11/2017(UTC) Posts: 82 Location: Midwest Was thanked: 9 time(s) in 7 post(s)
|
Sorry Jean, your genfit file won't find beta. It works fine when I download, but fails when I open on my pc.
nas
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: nasch Sorry Jean, your genfit file won't find beta. It works fine when I download, but fails when I open on my pc. 1. The work sheet attached finds the β's of the Verhulst logistic function, which corresponds to your collection. For the two parameters Weibull, we have to scale the Y data to [0..1] and the X data to something else. Your project up until now is just an undefined collection. I will attempt a pure Weibull fit. 2. Can't understand it works fine from download but fails on your PC. Like saying: moon is flat because she is flat as well as spherical because she is spherical. Just circular like Shrodinger cat ... hit the cat he is dead, hit again back live. Cheers ... Jean Stat Treasury_7 Weibull.sm (10kb) downloaded 10 time(s).
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: Jean Giraud I will attempt a pure Weibull fit. ... here is a typical two parameters Weibull Stat Treasury_7 Weibull.sm (19kb) downloaded 13 time(s).
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: Jean Giraud ... here is a typical two parameters Weibull ... too reactive for :Genfit Minimize. Thus: simply manual fit by the look. Stat Treasury_7 Weibull.sm (21kb) downloaded 21 time(s).
|
|
|
|
SMath Studio Forum
»
SMath Studio
»
Questions
»
Chain Failure Weibull Analysis
Forum Jump
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.