A multi-dimensional dynamic linear model for monitoring slaughter pig production

Dan Børge Jensen, Cecile Cornou, Nils Toft, Anders Kristensen

    Research output: Contribution to conferencePaperResearchpeer-review

    194 Downloads (Pure)

    Abstract

    Scientists and farmers still lack an efficient way to unify the large number of different types of data series, which are increasingly being generated in relation to automatic herd monitoring. Such a unifying model should be able to account for the correlations between the various types of data, resulting in a model which could potentially yield more information than can be gained from the individual components separately. Here we present such a model for monitoring slaughter pig production, in the form of a multivariate dynamic linear model. This model unifies three types of data (live weight, feed-and water consumption), measured at different levels of detail (individual pig and double-pen level) and with different observational frequencies (weekly and daily), using series collected for the Danish PigIT project. The presented three-dimensional model serves as a proof of concept, and it should be straightforward to expand it with additional data types.
    Original languageEnglish
    Publication date2015
    Number of pages8
    Publication statusPublished - 2015
    Event7th European Conference on Precision Livestock Farming (EC-PLF) - Milan, Italy
    Duration: 15 Sep 201518 Sep 2015
    Conference number: 7

    Conference

    Conference7th European Conference on Precision Livestock Farming (EC-PLF)
    Number7
    CountryItaly
    CityMilan
    Period15/09/201518/09/2015

    Keywords

    • Dynamic linear model
    • Information unification
    • Modeling
    • Monitoring
    • Pig production

    Fingerprint Dive into the research topics of 'A multi-dimensional dynamic linear model for monitoring slaughter pig production'. Together they form a unique fingerprint.

    Cite this