1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.
Dismiss Notice
You must be a registered member in order to post messages and view/download attached files in this forum.
Click here to register.

Issues with Minitab - Interval-Censored Data Set

Discussion in 'Reliability Analysis' started by Dan D, Apr 3, 2025.

Tags:
  1. Dan D

    Dan D Member

    Joined:
    Oct 30, 2024
    Messages:
    11
    Likes Received:
    0
    Trophy Points:
    1
    I am attempting to run Reliability Modeling in Minitab for the following data set.

    Sample ID Cyc_Start Cyc_End Target
    Sample01 30000 45000 15000
    Sample02 30000 45000 15000
    Sample03 30000 45000 15000
    Sample04 15000 30000 15000
    Sample05 15000 30000 15000
    Sample06 15000 30000 15000
    Sample07 15000 30000 15000
    Sample08 30000 45000 15000

    When I run the Stat>Reliability/Survival>Distribution Analysis (Arbitrary Censoring)>Distribution Plot ID, I get the following error. My Start variable is 'Cyc_Start", my End variable is 'Cyc_End', I have Use all distrubtions selected, and under Options, I have Least Squares selected.

    * ERROR * Least squares estimates could not be calculated for these data. There must be at
    least one distinct plot point on a probability plot for each estimated parameter.

    I then tried to do Distrubution Plot Overview using the 3-parameter Weibull as the Distribution model and got the same error.

    Can anyone provide some guidance as to what I am going wrong in Minitab?
     
    Dan D, Apr 3, 2025
    #1
  2. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    655
    Likes Received:
    562
    Trophy Points:
    92
    Location:
    Greater Milwaukee USA
    Minitab expects the data for arbitrary censoring to be formatted as follows:
    upload_2025-4-3_9-12-46.png

    I recommend trying to enter your data as shown:
    upload_2025-4-3_9-13-14.png
     
    Miner, Apr 3, 2025
    #2
    Dan D likes this.
  3. Dan D

    Dan D Member

    Joined:
    Oct 30, 2024
    Messages:
    11
    Likes Received:
    0
    Trophy Points:
    1
    Tried that under the Distribution Plot ID option...I got this error...

    * ERROR * Least squares estimates could not be calculated for these data. There must be at
    least one distinct plot point on a probability plot for each estimated parameter.

    Tried the Distribution Overview Plot option and got this error:

    * ERROR * The number of estimated parameters must be less than or equal to the number of
    distinct rows of data.

    upload_2025-4-3_10-40-20.png
     
    Dan D, Apr 3, 2025
    #3
  4. Dan D

    Dan D Member

    Joined:
    Oct 30, 2024
    Messages:
    11
    Likes Received:
    0
    Trophy Points:
    1
    upload_2025-4-3_10-41-52.png
     
    Dan D, Apr 3, 2025
    #4
  5. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    655
    Likes Received:
    562
    Trophy Points:
    92
    Location:
    Greater Milwaukee USA
    Try replacing the zero in row 1 with an *. If that doesn't work try deleting row 1.

    I'm working from memory because Minitab priced themselves out of our enterprise license market and we stopped using them several months ago.
     
    Miner, Apr 3, 2025
    #5
  6. Dan D

    Dan D Member

    Joined:
    Oct 30, 2024
    Messages:
    11
    Likes Received:
    0
    Trophy Points:
    1
    Alas, that didn't work either...I've tried finding Minitab instructional videos but I haven't been able to find any good quality options.
     
    Dan D, Apr 3, 2025
    #6
  7. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    655
    Likes Received:
    562
    Trophy Points:
    92
    Location:
    Greater Milwaukee USA
    I think I understand the issue. You only have two intervals that contain failures. That's like trying to fit a curve with only two points. You can't. You can only fit a straight line. Do you have any censored data? Units that have not yet failed but do have test hours racked up on them?
     
    Miner, Apr 3, 2025
    #7
  8. Dan D

    Dan D Member

    Joined:
    Oct 30, 2024
    Messages:
    11
    Likes Received:
    0
    Trophy Points:
    1
    That makes sense. No I don't...
     
    Dan D, Apr 3, 2025
    #8
  9. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    655
    Likes Received:
    562
    Trophy Points:
    92
    Location:
    Greater Milwaukee USA
    Do you have exact times of failure? Or, can you narrow the intervals such that you have the failures spread out over more intervals?
     
    Miner, Apr 3, 2025
    #9
  10. Dan D

    Dan D Member

    Joined:
    Oct 30, 2024
    Messages:
    11
    Likes Received:
    0
    Trophy Points:
    1
    I'm afraid not. The sampling interval was minimized due to resource constraints.
     
    Dan D, Apr 3, 2025
    #10
  11. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    655
    Likes Received:
    562
    Trophy Points:
    92
    Location:
    Greater Milwaukee USA
    I understand the resource constraints of having someone check periodically. Is the failure something that has to be verified in person or can it be sensed electronically such that it can record the time or cycles at failure? Are the failures "sudden death" failures without warning, or something that progresses and the progression can be measured, like the length of crack formation or the wear of a brake pad? If it can be measured, you can perform a degradation analysis.
     
    Miner, Apr 3, 2025
    #11