0
$\begingroup$

I have a question regarding the best way to represent survival data to predict the reliability of bike parts.

I have the manufacturing date of each bike. At the manufacturing data the bike is equipped with equipment that sends when a part breaks down. I do not have information on when a bike is repaired and back on the road.

╔═════════╦══════╦════════════╗
║ Bike ID ║ Time ║ PartNumber ║
╠═════════╬══════╬════════════╣
║       1 ║   10 ║         10 ║
║       1 ║  100 ║        120 ║
║       1 ║  110 ║         10 ║
║       1 ║  120 ║          1 ║
║       2 ║    0 ║          1 ║
║       2 ║   10 ║          3 ║
║       3 ║   50 ║        120 ║
║       3 ║   60 ║        140 ║
║       3 ║   70 ║          3 ║
╚═════════╩══════╩════════════╝

I am interested in the following analyses:

  1. Reliability of individual parts e.g. 10, 120;
  2. Competing risk for a bike break down; and,
  3. Prediction of what event will occur at a certain time.

I tend to use R.

$\endgroup$
6
  • $\begingroup$ Not really a competing risk model. Those assume that one event is a censoring process for the other evetns. $\endgroup$
    – DWin
    Sep 3, 2015 at 4:26
  • $\begingroup$ Could you elaborate? $\endgroup$
    – Stereo
    Sep 3, 2015 at 4:35
  • $\begingroup$ If you die of cancer you cannot later die of heart failure. $\endgroup$
    – DWin
    Sep 3, 2015 at 5:46
  • $\begingroup$ But what if I am interested in knowing whether someone is more likely to die of cancer than heart failure. $\endgroup$
    – Stereo
    Sep 4, 2015 at 2:21
  • $\begingroup$ The estimates of the rates of each will be biased. However, you may be able to rank the rates relative to each other. $\endgroup$
    – DWin
    Sep 4, 2015 at 2:59

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.