Abstract
With the growth of AI and data modelling, the old saying by George Fuechsel regarding data quality ‘Garbage in, garbage out’ holds more truth than ever. Data Scientists are learning the quality of their models depends on the quality of data. Data used by the Global Burden of Animal Diseases (GBADs) is available to modellers around the world, and the quality of the data provided is important as it is used in modelling disease, greenhouse gas emissions, and more. These are important topics, so the data given to the modellers must be investigated and checked for internal and external inconsistencies. The goal of this paper is to investigate data provided by GBADs to find inconsistencies in the data. Data quality was analysed using a five-year trailing average comparison, the interquartile range for the yearly rates of change, and observing outliers on a normal distribution for the yearly rates of change for livestock populations over time. The normal distribution and interquartile range analysis is an internal data analysis that can find outliers that indicate possible data inconsistencies. The five-year trailing average helps identify external data inconsistencies between sources. Using purpose-built data analysis tools and performing analysis on the data shows there are inconsistencies in the data. The consequences of these findings show that researchers need to be cognisant of the data they are using and need to perform their own analysis before they use it in their models as the data can show incorrect results.