A case study for the prediction of the number of pigs marketed and the risk analysis using Monte-Carlo simulation
2009
Unn, S.(Tokyo Univ. of Agriculture (Japan)) | Sugimoto, T. | Kurokawa, A. | Takahashi, H. | Akachi, K.
Today, because of the increase in operational sizes and asset values for livestock production, computers must effectively be utilized not only for data management but also for strategic risk analysis. In this paper, by using field data collected from 79 farrow-finish swine operations for one year, 13 productivity indices in the reproduction, which include the total number of pigs born per litter (TB) and the number of pigs weaned per sow per year (PWSY), and 10 growth productivity indices, including the annual death rate and the predicted number of pigs marketed were analyzed. Their appropriate distributions were first determined individually, and the goodness of fit was tested by Anderson-Darling test. By utilizing those fitted distributions and the Monte Carlo simulations, the sensitivity analysis revealed that TB had the largest positive effect (29.2%), followed by the largest negative effects of non-productive sow days (-28.7%) and piglet death rate during lactation (-28.7%) on PWSY. Additionally, the number of stillborn pigs per litter (-9.6%) and the lactation length (-3.2%) were shown to have comparable effects. Risk analysis for the number of pigs marketed and the total sales under the actual cases showed the relationship between the sales and its reliability, and revealed the effectiveness in actual usage for the decision-making.
显示更多 [+] 显示较少 [-]