An algorithm to process, clean and aggregate the data from IVOG® electronic feeding stations

This combination of two RMarkdown reports describes a process of processing, cleaning and aggregating the data of IVOG® electronic feeding stations (Hokofarm group, the Netherlands). Along the way, the algorithms create visualisations and quantifications of the data, the choices made and their effects. In addition, different datasets are created and exported for future use. As all the code is provided and RMarkdown files can be easily reproduced, these algorithms can be used to clean your data, aggregate it to different levels, and gain more insight into your data's patterns and quality. In some cases and with some adaptations, the algorithm could also be applied to other types of feeding stations and sensors, for more information on this we refer to the README file. The algorithm covers eight main parts, split across two reports: 1) Pre-processing to put the data in the right format; 2) Assessing data completeness and visualisation of its pre-cleaning quality; 3) Cleaning of the data using pre-set rules; 4) Visualisation and summary of the cleaning results; 5) Aggregation of the data to different time and subject levels; 6) Calculation of a meal criterion using a three-part probability density function; 7) Application of the meal criterion to aggregate the data further to the meal level; and 8) Visualisation of the meal-level data. More information on different parts of the algorithm and on preparatory steps to take before the algorithm can run can be found shortly in the first chunk of each RMarkdown file ('setup') and more elaborately in the README file. A description of all used variables can also be found in the README file. For the most effective use of the algorithm, we advice to start by reading the README file.本组合式两份R标记文档（RMarkdown）阐述了针对IVOG®电子饲喂站（荷兰霍科农场集团（Hokofarm Group）出品）的数据处理、清洗与聚合流程。在此流程中，算法可生成数据的可视化结果与量化分析，并阐释算法参数选择及其带来的影响。此外，本流程还会生成并导出多份数据集以供后续研究使用。 由于所有代码均已开源，且R标记文档可便捷复现，用户可利用本算法完成数据清洗、多维度聚合，并深入挖掘数据的模式与质量特征。在部分场景下，经适当适配后，本算法亦可应用于其他类型的饲喂站与传感器设备，相关细节请参阅README文件。 本算法共包含八大核心模块，分布于两份文档之中： 1. 预处理：将数据调整至标准格式； 2. 数据完整性评估与清洗前质量可视化； 3. 基于预设规则的数据清洗； 4. 清洗结果的可视化与汇总分析； 5. 将数据按不同时间维度与个体维度进行聚合； 6. 基于三分段概率密度函数计算采食判定标准； 7. 应用采食判定标准将数据进一步聚合至采食事件层级； 8. 采食事件层级数据的可视化。 关于算法各模块的详细说明，以及算法运行前所需完成的准备步骤，可在每份R标记文档的首个代码块（setup）中快速查阅，更详尽的内容则可参阅README文件。所有已使用变量的说明亦可在README文件中查阅。为最大化发挥本算法的效用，我们建议用户首先阅读README文件。