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.

Analyzing Data Set with Right- and Left-Censored Data

Discussion in 'Reliability Analysis' started by Dan D, Oct 30, 2024.

  1. Dan D

    Dan D Member

    Joined:
    Oct 30, 2024
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    1
    Hello all,

    I am trying to analyze some data sets for reliability in Minitab. Several of these data sets have right-censored and left-censored data and I cannot find a way to analyze these data sets using the Distribution Analysis (Arbitrary Censoring) function within Minitab. Can anyone help me?
     
  2. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    627
    Likes Received:
    543
    Trophy Points:
    92
    Location:
    Greater Milwaukee USA
    I can help you if you can attach your data.
     
  3. Dan D

    Dan D Member

    Joined:
    Oct 30, 2024
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    1
    Thank you. See attached. I have found a lot of information on how to process right-censored data or left-censored data in Minitab, but I have had no luck finding left- and right-censored data processing examples.
     

    Attached File(s): 1. Scan for viruses before using. 2. Report any 'bad' files by reporting this post. 3. Use at your own Risk.:

  4. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    627
    Likes Received:
    543
    Trophy Points:
    92
    Location:
    Greater Milwaukee USA
    Please explain what YOU mean by Cyc Start and Cyc End. For example, on Test_2 Cyc Start is 1000 (cycles?). If you know that it already has 1000 cycles on it, it's not left censored. Left censoring is usually associated with time rather than cycles. You know a date when the event occurred such as the product failed in the field, but you don't know when that product was made or put into service. Or, am I misinterpreting this?

    upload_2024-10-31_10-54-55.png

    I may be wrong, but this is how I interpret your data.
    upload_2024-10-31_11-5-39.png
     
    Last edited: Oct 31, 2024
  5. Dan D

    Dan D Member

    Joined:
    Oct 30, 2024
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    1
    Your interpretation is correct. The "failed" readings failed between the Cyc Start and Cyc End values, hence why they are left-censored as I don't know the exact point of failure. The "censored" values are right censored and represent samples that are still actively running.
     
  6. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    627
    Likes Received:
    543
    Trophy Points:
    92
    Location:
    Greater Milwaukee USA
    They would be interval censored then, not left censored. Test_9 failed somewhere between 100,000 and 200,000 cycles. I will work on the analysis tomorrow.
     
  7. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    627
    Likes Received:
    543
    Trophy Points:
    92
    Location:
    Greater Milwaukee USA
    I recommend that you read this help article from Minitab: Enter your data for Distribution ID Plot (Arbitrary Censoring). It will explain how to enter your data for each type of censoring. You have interval censoring, which means you do not have an exact time of failure, just that it occurred within a known interval. The graphic below shows how you should enter your data. Columns 2 & 3 are all Minitab needs.

    When you run your analysis, check under options for the estimation method. You have a relatively small number of failures, so you should use the Least Squares estimation method until you get at least 30 failures. The default method is Maximum Likelihood, so make sure you change it. I found the 3-parameter Weibull distribution to be a good fit.

    upload_2024-11-1_7-33-7.png

    upload_2024-11-1_7-39-32.png
     
    Dan D likes this.
  8. Dan D

    Dan D Member

    Joined:
    Oct 30, 2024
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    1
    Thank you, I'll give that a try!
     
  9. Dan D

    Dan D Member

    Joined:
    Oct 30, 2024
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    1
    @Miner When running the Distribution ID Plot on this data, I'm getting this warning. I can't seem to find a satisfactory explanation online as the implications of this warning. Do you have any insight?
    * WARNING * LSXY algorithm has not converged to optimal threshold parameter(s).
     
  10. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    627
    Likes Received:
    543
    Trophy Points:
    92
    Location:
    Greater Milwaukee USA
    I run into that warning almost every time. Please read Minitab's explanation of the cause and how to deal with it in this help article. It has to do with the default limit of 20 iterations before it converges on a solution. I have found that even with the error, it doesn't affect the decision.
     
    Atul Khandekar and Dan D like this.