I wish JMP supported Weibull analysis with left-truncated datasets. While it handles left censoring, it doesn’t yet address left truncation, even though the math is well-established. This is an important real-world need in reliability analysis, and adding it would make the Weibull platform better
This is a very real problem: For Product A, we have the complete installed population. In year 9, we discovered a new failure mode and began collecting data from that point forward. Any failures that may have occurred between years 1–9 are unknown and unobserved. This creates a left-truncated dataset, because we only see units that survived long enough to be observed starting in year 10.
This is fundamentally different from left censoring. With left censoring, you know a failure occurred before a certain time, but not the exact moment — the unit is still included in the dataset with partial information. With left truncation, early failures never appear in the dataset at all, and the analysis must condition on survival past the truncation point.
To accurately model the Weibull distribution for this failure mode, we need JMP’s platform to support left truncation, since it reflects real-world reliability scenarios where monitoring starts mid-life.
Currently to solve this problem you have to use Python, Reliasoft or other platform. The Math is well established and there is no reason JMP should not be able to handle it
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.