Creating a model to detect dairy cattle farms with poor welfare using a national database

C Krug, MJ Haskell, TP Nunes, G Stilwell

Research output: Contribution to journalArticle

4 Citations (Scopus)
2 Downloads (Pure)

Abstract

The objective of this study was to determine whether dairy farms with poor cow welfare could be identified using a national database for bovine identification and registration that monitors cattle deaths and movements. The welfare of dairy cattle was assessed using the Welfare Quality® protocol (WQ) on 24 Portuguese dairy farms and on 1,930 animals. Five farms were classified as having poor welfare and the other 19 were classified as having good welfare. Fourteen million records from the national cattle database were analysed to identify potential welfare indicators for dairy farms. Fifteen potential national welfare indicators were calculated based on that database, and the link between the results on the WQ evaluation and the national cattle database was made using the identification code of each farm. Within the potential national welfare indicators, only two were significantly different between farms with good welfare and poor welfare, 'proportion of on-farm deaths' (p < 0.01) and 'female/male birth ratio' (p < 0.05). To determine whether the database welfare indicators could be used to distinguish farms with good welfare from farms with poor welfare, we created a model using the classifier J48 of Waikato Environment for Knowledge Analysis. The model was a decision tree based on two variables, 'proportion of on-farm deaths' and 'calvingto- calving interval', and it was able to correctly identify 70% and 79% of the farms classified as having poor and good welfare, respectively. The national cattle database analysis could be useful in helping official veterinary services in detecting farms that have poor welfare and also in determining which welfare indicators are poor on each particular farm.
Original languageEnglish
Pages (from-to)280-286
JournalPreventive Veterinary Medicine
Volume122
Issue number3
DOIs
Publication statusFirst published - 2015

Fingerprint

dairy farming
Databases
farms
cattle
Farms
death
calving interval
dairy cattle
cows
Parturition

Bibliographical note

1023365

Keywords

  • Animal welfare
  • Dairy cattle
  • National cattle database
  • Welfare quality

Cite this

@article{59b06f3aae874e03bf7a97cd9ab483a0,
title = "Creating a model to detect dairy cattle farms with poor welfare using a national database",
abstract = "The objective of this study was to determine whether dairy farms with poor cow welfare could be identified using a national database for bovine identification and registration that monitors cattle deaths and movements. The welfare of dairy cattle was assessed using the Welfare Quality{\circledR} protocol (WQ) on 24 Portuguese dairy farms and on 1,930 animals. Five farms were classified as having poor welfare and the other 19 were classified as having good welfare. Fourteen million records from the national cattle database were analysed to identify potential welfare indicators for dairy farms. Fifteen potential national welfare indicators were calculated based on that database, and the link between the results on the WQ evaluation and the national cattle database was made using the identification code of each farm. Within the potential national welfare indicators, only two were significantly different between farms with good welfare and poor welfare, 'proportion of on-farm deaths' (p < 0.01) and 'female/male birth ratio' (p < 0.05). To determine whether the database welfare indicators could be used to distinguish farms with good welfare from farms with poor welfare, we created a model using the classifier J48 of Waikato Environment for Knowledge Analysis. The model was a decision tree based on two variables, 'proportion of on-farm deaths' and 'calvingto- calving interval', and it was able to correctly identify 70{\%} and 79{\%} of the farms classified as having poor and good welfare, respectively. The national cattle database analysis could be useful in helping official veterinary services in detecting farms that have poor welfare and also in determining which welfare indicators are poor on each particular farm.",
keywords = "Animal welfare, Dairy cattle, National cattle database, Welfare quality",
author = "C Krug and MJ Haskell and TP Nunes and G Stilwell",
note = "1023365",
year = "2015",
doi = "10.1016/j.prevetmed.2015.10.014",
language = "English",
volume = "122",
pages = "280--286",
journal = "Preventive Veterinary Medicine",
issn = "0167-5877",
publisher = "Elsevier",
number = "3",

}

Creating a model to detect dairy cattle farms with poor welfare using a national database. / Krug, C; Haskell, MJ; Nunes, TP; Stilwell, G.

In: Preventive Veterinary Medicine, Vol. 122, No. 3, 2015, p. 280-286.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Creating a model to detect dairy cattle farms with poor welfare using a national database

AU - Krug, C

AU - Haskell, MJ

AU - Nunes, TP

AU - Stilwell, G

N1 - 1023365

PY - 2015

Y1 - 2015

N2 - The objective of this study was to determine whether dairy farms with poor cow welfare could be identified using a national database for bovine identification and registration that monitors cattle deaths and movements. The welfare of dairy cattle was assessed using the Welfare Quality® protocol (WQ) on 24 Portuguese dairy farms and on 1,930 animals. Five farms were classified as having poor welfare and the other 19 were classified as having good welfare. Fourteen million records from the national cattle database were analysed to identify potential welfare indicators for dairy farms. Fifteen potential national welfare indicators were calculated based on that database, and the link between the results on the WQ evaluation and the national cattle database was made using the identification code of each farm. Within the potential national welfare indicators, only two were significantly different between farms with good welfare and poor welfare, 'proportion of on-farm deaths' (p < 0.01) and 'female/male birth ratio' (p < 0.05). To determine whether the database welfare indicators could be used to distinguish farms with good welfare from farms with poor welfare, we created a model using the classifier J48 of Waikato Environment for Knowledge Analysis. The model was a decision tree based on two variables, 'proportion of on-farm deaths' and 'calvingto- calving interval', and it was able to correctly identify 70% and 79% of the farms classified as having poor and good welfare, respectively. The national cattle database analysis could be useful in helping official veterinary services in detecting farms that have poor welfare and also in determining which welfare indicators are poor on each particular farm.

AB - The objective of this study was to determine whether dairy farms with poor cow welfare could be identified using a national database for bovine identification and registration that monitors cattle deaths and movements. The welfare of dairy cattle was assessed using the Welfare Quality® protocol (WQ) on 24 Portuguese dairy farms and on 1,930 animals. Five farms were classified as having poor welfare and the other 19 were classified as having good welfare. Fourteen million records from the national cattle database were analysed to identify potential welfare indicators for dairy farms. Fifteen potential national welfare indicators were calculated based on that database, and the link between the results on the WQ evaluation and the national cattle database was made using the identification code of each farm. Within the potential national welfare indicators, only two were significantly different between farms with good welfare and poor welfare, 'proportion of on-farm deaths' (p < 0.01) and 'female/male birth ratio' (p < 0.05). To determine whether the database welfare indicators could be used to distinguish farms with good welfare from farms with poor welfare, we created a model using the classifier J48 of Waikato Environment for Knowledge Analysis. The model was a decision tree based on two variables, 'proportion of on-farm deaths' and 'calvingto- calving interval', and it was able to correctly identify 70% and 79% of the farms classified as having poor and good welfare, respectively. The national cattle database analysis could be useful in helping official veterinary services in detecting farms that have poor welfare and also in determining which welfare indicators are poor on each particular farm.

KW - Animal welfare

KW - Dairy cattle

KW - National cattle database

KW - Welfare quality

U2 - 10.1016/j.prevetmed.2015.10.014

DO - 10.1016/j.prevetmed.2015.10.014

M3 - Article

VL - 122

SP - 280

EP - 286

JO - Preventive Veterinary Medicine

JF - Preventive Veterinary Medicine

SN - 0167-5877

IS - 3

ER -