Max Ghenis

Why was this closed?

See https://stats.stackexchange.com/a/107822/19676

To clarify, in openfisca-uk we don't have separate uprating factors for each variable, we assign variables one of a handful of uprating factors. I think @nikhilwoodruff just showed it paired...

>In order to help prioritise and avoid resource wastage, could you please share any deployment constraints you have / know of? No constraints, though there could be general performance improvements....

Since Python 3.9 has issues (https://github.com/openfisca/openfisca-core/issues/1035#issuecomment-913693451), I shrunk this issue down to Python 3.8. Python 3.8 support would be especially useful since it's the default conda Python version, and that's...

numpy 1.17.0 is failing to install on Python 3.8. I don't understand why it's starting at numpy 1.19.0 to retrieve the most recent three versions, rather than 1.22.0, which is...

One example of this is https://github.com/PolicyEngine/openfisca-us/pull/387, where we're creating test cases for a benefit program from [this third party calculator](https://bplc.cssny.org/benefit_tools/snap_calculator).

Here's how [Tibshirani et al (2011)](https://www.stat.cmu.edu/~ryantibs/papers/neariso.pdf) describe isotonic regression:  The Pool Adjacent Violators Algorithm (PAVA) solves this; I found [this description](https://stat.fandom.com/wiki/Isotonic_regression) of PAVA helpful.

@rickecon could you share sample data to work through? I only cited the Tibshirani paper for the description of isotonic regression, agree that their approach is unsuitable.

Here's R code for multiple isotonic regression, written by Meyer, an apparent expert on isotonic regression: https://www.stat.colostate.edu/~meyer/multipleiso.htm [Meyer (2013)](https://www.tandfonline.com/doi/pdf/10.1080/03610918.2012.659820) shows this relevant diagram:  Meyer is also a coauthor on...