Grow-Finish Data: An Untapped Resource
Use your own data to make informed decisions.
By Stephanie Rutten-Ramos
From weekly performance monitors in the sow herd to group closeouts in finishing, pork producers are accustomed to looking at data. Targets are established, performance is compared and improvements are pursued.
When it comes to improvement insights, however, nursery and finishing data can be challenging. Unlike sow information, which is tracked to the level of the individual, nursery and finishing data largely rely on measurements summarized to the level of the group - average ages and weights of pigs in and out, for example. Further complicating matters is the fact that performance expectations change as pigs grow, since bigger pigs gain faster but less efficiently.
This can make group comparisons difficult and misleading. Although Excel makes it easy to generate a two-variable scatterplot with added "trendline," few relationships in pig production are that simple. Yet understanding the relationships and subtleties of the data is critical for improvement.
Some parameters vary by season (for example, feed consumption is better with new-crop corn). Other parameters, like mortality rate, may vary with the type of pigs placed (i.e., gilts/barrows/ mixed groups).
In the pursuit for improvement, distinguishing which variables are related to outcomes is important for two reasons. First, interventions can have great benefit if they can be applied at the right time - the difference between best management practices (fire prevention) and pig treatments needed for not incorporating those practices (fire fighting). And second, differentiating between the potential for local improvement and system-wide improvement allows a system to efficiently use its resources.
Remember that no two systems are identical. There are differences in herd health levels, maturity and genetic base. There are differences in regions, facilities and feedstuffs. Even diet formulations within a system may vary according to ingredient costs and season of the year. That being said, the remainder of this space will be used to explore some relationships between performance measures and group descriptors for a set of finishing data.
A look at finishing data
This dataset reflects one year of production from a mixed-sex finishing system supplied by three different pig sources (89 groups). Their outcomes of interest are mortality rate, average daily gain (excluding live weight gain by deaths), and carcass weight/pig. The factors of interest are source [Flow], the time of year in which pigs were placed [StartQtr], and Pigs in Average Weight [StartWt]. For analytical purposes, Pigs in Average Weight was ranked low to high within Flow and divided into three categories: Light, Medium, and Heavy.
Figure 1 is a histogram of mortality rates from the dataset. Figures 2-4 depict relationships between mortality rate and Pigs In Average Weight, Flow, and both Flow and Pigs In Average Weight category. By simultaneously considering both Flow and the Pigs In Average Weight category, this system finds not only that Flow B has a higher mortality rate, but also that mortality rates are significantly higher in the lightest third of groups placed. With this knowledge, the system can develop strategies to reduce mortality with interventions specific to Flow B and, especially, Flow B lightweight groups.
Average daly grain
Figure 5 is a histogram of average daily gains. Figures 6 and 7 depict relationships between average daily gain and Flow, and average daily gain and StartQtr. For this dataset, average daily gains differ across both StartQtr and Flow, but not Pigs in Average Weight. There was no interaction between Flow and StartQtr, indicating a seasonal gain pattern consistent across all sources. With this knowledge, the system can target interventions both by Flow and according to the time of year when pigs are placed.
Best practices yield results
Even if an industry isn't operating on thin margins, it only makes good business sense to employ practices that yield results. Many systems have a standard set of best practices, yet differences in system inputs (pig size, health, feedstuffs) and facilities (barn style, Feeder types, etc.) may warrant interventions. When a system can identify characteristics of groups less likely to achieve production targets, it can direct interventions only to those groups most likely to benefit. If, however, there are no real differences in group performance across the descriptive variables, then system wide interventions should be considered.