Heifer Inventory Prediction

[paradox3206]

Despite similar regional production, diet nutrients, and cow factors, RuFaS predicted variable calf and heifer inventories (see below). Because of this issue, only RuFaS outputs specifically for lactating cows could be compared for our work. I am confused as to why calf and heifer inventories would change simply with minor diet ingredient changes and no nutrition or management changes.

	GL31BP	GL42BP	GL45BP	NE0BP	NE23BP	NE41BP	NE48BP
Calves_ver_agg	8	9	9	10	8	10	10
Heifers_hor_ver_agg	94	89	77	85	83	81	85
Dry cows_ver_agg	13	12	12	13	12	12	12
Milking cows_ver_agg	88	88	88	87	88	88	88

	W0BP	W37BP	W48BP	W51BP
Calves_ver_agg	7	8	8	9
Heifers_hor_ver_agg	90	85	85	89
Dry cows_ver_agg	12	13	13	12
Milking cows_ver_agg	89	88	88	88

[KFosterReed]

Another good question @paradox3206 :)

@JoeWaddell mentioned this in a meeting about the randomness in the animal module. Because the herd structure of a single simulation is the result of many random events for individual animals and ration formulation, if you change the number of random draws in a scenario by changing the number of feeds or the number of times the ration formulation algorithm iterates over a diet, it can have a snowball effect on the outcomes of the events for the animal event random draws and then the total herd structure. Because of this for inference on outcomes that are impacted by the animal module, we recommend averaging 10 scenarios. Were the above results from an average of multiple simulations?

I know it is not ideal for a change in diet to affect the herd this way merely due to randomness rather than some expected biological outcome and, somewhat serendipitously, we started to think about ways to reduce the interdependence between random processes that are not intended to be so dependent (like an animal conception success and ration optimization). @allisterakun opened issue #3268 just yesterday (citing your use case!) as a start to making the randomness easier to manage.

[paradox3206]

Thanks for your reply. To answer your question, the data provided was the result of one simulation. So, it makes sense that averaging 10 scenarios would reduce variation in heifer inventory prediction. Similar to my issue with DMI prediction, both researchers and dairy farms cannot make accurate whole-farm comparisons of dietary strategies if predicted animal numbers vary. From: Kristan ***@***.***> Sent: Friday, September 11, 2026 4:53 PM To: RuminantFarmSystems/RuFaS ***@***.***> Cc: paradox3206 ***@***.***>; Mention ***@***.***> Subject: Re: [RuminantFarmSystems/RuFaS] Heifer Inventory Prediction (Discussion #3280) Another good question @paradox3206 <https://github.com/paradox3206> :) @JoeWaddell <https://github.com/JoeWaddell> mentioned this in a meeting about the randomness in the animal module. Because the herd structure of a single simulation is the result of many random events for individual animals and ration formulation, if you change the number of random draws in a scenario by changing the number of feeds or the number of times the ration formulation algorithm iterates over a diet, it can have a snowball effect on the outcomes of the events for the animal event random draws and then the total herd structure. Because of this for inference on outcomes that are impacted by the animal module, we recommend averaging 10 scenarios. Were the above results from an average of multiple simulations? I know it is not ideal for a change in diet to affect the herd this way merely due to randomness rather than some expected biological outcome and, somewhat serendipitously, we started to think about ways to reduce the interdependence between random processes that are not intended to be so dependent (like an animal conception success and ration optimization). @allisterakun <https://github.com/allisterakun> opened issue #3268 <#3268> just yesterday (citing your use case!) as a start to making the randomness easier to manage. — Reply to this email directly, view it on GitHub <#3280?email_source=notifications&email_token=B7P5GD2OTZDNIINBBPKETCT5ORQ2PA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBUGA3DGNJXUZZGKYLTN5XKO3LFNZ2GS33OUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y#discussioncomment-18406357> , or unsubscribe <https://github.com/notifications/unsubscribe-auth/B7P5GD6HFFR4C7DYJ3HW4UL5ORQ2PAVCNFSNUABIKJSXA33TNF2G64TZHMYTIMBUG44TAOJRHNCGS43DOVZXG2LPNY5TCMBXHE3TGNZVUF3AE> . Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS <https://github.com/notifications/mobile/ios/B7P5GD6N3OHL666EUESH46D5ORQ2PA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBUGA3DGNJXUZZGKYLTN5XKO3LFNZ2GS33OUVSXMZLOOSVGM33PORSXEX3JN5ZQ> and Android <https://github.com/notifications/mobile/android/B7P5GD5NS7YG5RNUEMHVHED5ORQ2PA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBUGA3DGNJXUZZGKYLTN5XKO3LFNZ2GS33OUVSXMZLOOSXGM33PORSXEX3BNZSHE33JMQ> . Download it today! You are receiving this because you were mentioned.Message ID: ***@***.***>