BACKGROUND Payment programs based on milk quality (PPBMQ) are used in several countries around the world as an incentive to improve milk quality. One of the principal milk parameters used in such programs is the bulk tank somatic cell count (BTSCC). In this study, using data from an average of 37,000 farms per month in Brazil where milk was analyzed, BTSCC data were divided into different payment classes based on milk quality. Then, descriptive and graphical analyses were performed. The probability of a change to a worse payment class was calculated, future BTSCC values were predicted using time series models, and financial losses due to the failure to reach the maximum bonus for the payment based on milk quality were simulated. RESULTS In Brazil, the mean BTSCC has remained high in recent years, without a tendency to improve. The probability of changing to a worse payment class was strongly affected by both the BTSCC average and BTSCC standard deviation for classes 1 and 2 (1000-200,000 and 201,000-400,000 cells/mL, respectively) and only by the BTSCC average for classes 3 and 4 (401,000-500,000 and 501,000-800,000 cells/mL, respectively). The time series models indicated that at some point in the year, farms would not remain in their current class and would accrue financial losses due to payments based on milk quality. CONCLUSION The BTSCC for Brazilian dairy farms has not recently improved. The probability of a class change to a worse class is a metric that can aid in decision-making and stimulate farmers to improve milk quality. A time series model can be used to predict the future value of the BTSCC, making it possible to estimate financial losses and to show, moreover, that financial losses occur in all classes of the PPBMQ because the farmers do not remain in the best payment class in all months.

中文翻译：

---

巴西乳牛群散装罐体细胞计数的逐月变化和预测，以及其对牛奶质量的影响。

背景技术基于牛奶质量的支付程序（PPBMQ）在世界上的多个国家中被用来激励人们改善牛奶质量。在此类程序中使用的主要牛奶参数之一是大罐体细胞计数（BTSCC）。在这项研究中，使用来自巴西平均每月37,000个农场进行牛奶分析的数据，BTSCC数据根据牛奶质量分为不同的付款类别。然后，进行描述性和图形分析。计算了更改为更差的付款类别的可能性，使用时间序列模型预测了将来的BTSCC值，并模拟了由于未能达到基于牛奶质量的最高付款奖金而造成的财务损失。结果在巴西，近年来平均BTSCC保持较高水平，而没有改善的趋势。更改为较差付款类别的可能性受BTSCC平均值和1类和2类BTSCC标准偏差（分别为1000-200,000和201,000-400,000个细胞/ mL）的强烈影响，而仅受3类和2类BTSCC平均值的影响。 4（分别为401,000-500,000和501,000-800,000细胞/ mL）。时间序列模型表明，在一年中的某个时候，农场将不再保持当前的水平，并且由于基于牛奶质量的付款而将产生经济损失。结论巴西奶牛场的BTSCC最近没有改善。等级改变为较差等级的可能性是可以帮助决策并刺激奶农提高牛奶质量的指标。时间序列模型可用于预测BTSCC的未来价值，从而可以估算财务损失并显示