search

IvanWilli / WPP_indirect_estimates

Indirect estimates on adult mortality based on methods: intercensal survival, siblings, orphanhood and widowhood.
0 stars 0 forks source link

Couple more questions on HIV functions #2

Open TTRustam opened 1 year ago

TTRustam commented 1 year ago

Hi Ivan. A couple more questions! First are about the predictNQX and error.svd functions. Can we get some information on them to add to the documentation? At the minimum, who is the author a very short description? That would be great! I also have 2 questions regarding these functions: 1) These functions refer to the object called mods.r that is supposed to be a matrix or a data.frame, but I do not see the definition of this object anywhere in the code. Am I missing something? 2) The predictNQX calls for a library rms but I did not find any hard-coded functions to be used from this library. Maybe this lib changes the methods or something like that? Can you update me on this? And finally a question regarding the rounding. You routinely use the rounding here and there in the code and sometimes I see, that it is necessary for life-table matching. My question is, is it ALWAYS used for matching the args or something similar OR sometimes you do it for unit tests to compare the results with already existing solutions or something similar? Thanks!

IvanWilli commented 1 year ago

Hi Rustam, Thanks! Actually my colleagues are deciding if include or not the HIV pattern, so please dont pay attention to those functions by the moment. We used it for our estimates but seems more is needed to get the final version on this. If they decide to leave it out (is more probable), I would do a commit with changes, sorry for the problems. The same function is used in other methods (orphanhood, siblings, widowhood), so this will save us time after. Yes, can be both. The rounding is for dates and e0 input mainly, that should be match-able with some mlt function. Otherwise, in "replicate" code, this is useful for testing (equal with 1 decimal) (UN Manual for example were done in 1980´s probably with pencil). Best,

TTRustam commented 1 year ago

Thanks for the clarifications! Highly appreciate. Then we will disregard the HIV functions for now! yeah, I thought about unit tests the moment I saw the rounding in the code. Thanks! :-) Best,

IvanWilli commented 1 year ago

Hi Rustam! We need to remove hiv pattern options. Do you need I make a commit with the function cleaned HIV functionalities? Best,

TTRustam commented 1 year ago

Hola Ivan! Nice to hear from you. Well, you can make a commit, yes, but alternatively, I can do it myself if you will tell me what exactly we want to remove to be extra sure)). Like which functions, or arguments etc.

IvanWilli commented 1 year ago

Basically arguments "HIV_prev" and "HIV_art" are not needed, Also removing lines 76:79, the condition on line 150 (to build an hiv family), and line 319 is unnecsary asking is.HIV Let me know if some other issue emerges Thanks!

TTRustam commented 1 year ago

Got it! Thanks a lot, Ivan. We will be in touch. Take care!

IvanWilli commented 1 year ago

Hi Rustam! How are you? How this is going? Do you need some help with anything? Best,

TTRustam commented 1 year ago

Hey Ivan! How are you? Hope everything is going fine. Sorry, was preoccupied with paper submission). Well actually yes, I was wondering if you could take a look at the reformated code. To eagle eye it so to speak for potential drawbacks. Can you do that? You can find the code here I`m not talking about thorough tests or something, just a general look over over it. Thanks!