AUTHOR=Martelli Francesco , Giacomozzi Claudia , Fadda Antonello , Frazzoli Chiara
TITLE=Understanding Seasonal Changes to Improve Good Practices in Livestock Management
JOURNAL=Frontiers in Public Health
VOLUME=6
YEAR=2018
URL=https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2018.00175
DOI=10.3389/fpubh.2018.00175
ISSN=2296-2565
ABSTRACT=Background and Aim: Food quality control techniques based on process control methods are increasingly adopted in livestock production systems to fulfill increasing market's expectations toward competitiveness and issues linked to One Health pillars (environment, animal, and human health). Control Charts allow monitoring and systematic investigation of sources of variability in dairy production parameters. These parameters, however, may be affected by seasonal variations that render impractical, biased or ineffective the use statistical control charts. A possible approach to this problem is to adapt seasonal adjustment methods used for the analysis of economic and demographic seasonal time series. The aim of the present work is to evaluate a seasonal decomposition technique called X-11 on milk parameters routinely collected also in small farms (fat, protein, and lactose content, solids-not-fat, freezing point, somatic cell count, total bacterial count) and to test the efficacy of different seasonal removal methods to improve the effectiveness of statistical control charting.
Method: Data collection was carried out for 3 years on routinely monitored bulk tank milk parameters of a small farm. Seasonality presence was statistically assessed on milk parameters and, for those parameters showing seasonality, control charts for individuals were applied on raw data, on X-11 seasonally adjusted data, and on data smoothed with a symmetric moving average filter. Correlation of seasonally influenced parameters with daily mean temperature was investigated.
Results: Presence of seasonality in milk parameters was statistically assessed for fat, protein, and solids-non-fat components. The X-11 seasonally-adjusted control charts showed a reduced number of violations (false alarms) with respect to non-seasonally adjusted control chart (from 5 to 1 violation for fat, from 17 to 1 violation for protein, and from 9 to none violation for solids-non-fat.). This result was achieved despite stricter control chart limits: with respect to raw data charts, the interval of control chart allowed variation (UCL–LCL) was reduced by 43% for fat, by 33.1% for protein, and by 14.3% for solids-not-fat.
Conclusions: X-11 deseasonalization of routinely collected milk parameters was found to be an effective method to improve control chart application effectiveness in farms and milk collecting centers.