DevelopmentConstant drops the tail factor

[kennethshsu]

Are you on the latest chainladder version?

• Yes, this bug occurs on the latest version.

Describe the bug in words

The DevelompentConstant() discards the last CDF to ultimate when the last CDF is not 1.0 (it sets it to 1.0).

How can the bug be reproduced?

reported_patterns = {
    12: 4.0, 24: 2.9, 36: 1.8, 48: 1.4, 60: 1.2, 72: 1.1, 84: 1.03, 96: 1.02, 108: 1.005,
}
auto_bi = cl.load_sample("friedland_auto_bi_insurer")
reported_BI_claim = cl.DevelopmentConstant(
    patterns=reported_patterns, style="cdf"
).fit_transform(auto_bi["Reported Claims"])
reported_BI_claim.cdf_

What is the expected behavior?

The 108-Ult factor of 1.005 should not be discarded.

[henrydingliu]

What is the expected behavior?
The 108-Ult factor of 1.005 is discarded.

Do you mean the opposite?

[kennethshsu]

Oh ya lol

[henrydingliu]

why does the example in the doc work, i.e. not drop the non-1 factor at the end?

[kennethshsu]

why does the example in the doc work, i.e. not drop the non-1 factor at the end?

Because in this example the pattern in DevelopmentConstant() doesn't actually go beyond the tail of the triangle. If you were to add another pattern 84: 1.01 or something like that it'll be discarded.

[henrydingliu]

ah, it's the fit_transform. i guess the package expects you to chain a developmentconstant with a tailconstant, which can be pain.

what about the regular development? like if you fit on a 10x10 triangle and then transform on a 9x9 triangle, do you get a similar behavior?