An Application of Information Granules to Detect Anomalies in COVID-19 Reports

The COVID-19 pandemic has affected almost every aspect of life. The patterns of interpersonal contacts, the ways of doing business and the methods of school education have changed. A significant part of worldwide business has migrated to the virtual world, and the global supply chains have been disrupted. On the other hand, this new situation created opportunities for a much faster development of some areas of business and science. For example, the observation and analysis of pandemic data has contributed to the development of new techniques for effective mathematical forecasting. It is worth noting that during a pandemic most political and economic decisions are based on official data on the number of new infections at the country level. Therefore, the quality of this data is very important for making difficult decisions, such as implementing new restrictions. In this study, we will focus on the problem of pandemic data quality and present a novel anomaly detection method based on information granules. In numerical experiments, data from several European countries were compared. The selection of data for analysis was based on the following information: the movement of people between countries, similar quality of medical care and the sanitary standards. An appropriate adaptation of the author's anomaly detection method based on information granules allowed to identify potential anomalies in daily COVID reports. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

The Concept of Granular Representation of the Information Potential of Variables

With the advent of research into Granular Computing, in particular information granules, the way of thinking about data has changed gradually. Researchers and practitioners do not consider only their specific properties, but also try to look at the data in a more general way, closer to the way people think. This kind of knowledge representation is expressed particularly in approaches based on linguistic modeling or fuzzy techniques such as fuzzy clustering, but also newer approaches related to the explanation of how artificial intelligence works on these data (so-called explainable artificial intelligence). Therefore, especially important from the point of view of the methodology of data research is an attempt to understand their potential as information granules. Such a kind of approach to data presentation and analysis may introduce considerations of a higher, more general level of abstraction, while at the same time reliably describing the network of relationships between the data and the observed information granules. In this study, we tackle this topic with particular emphasis on the problem of choosing a predictive model. In a series of numerical experiments based on both artificially generated data, ecological data on changes in bird arrival dates in the context of climate change, and COVID-19 infections data we demonstrate the effectiveness of the proposed approach built with a novel application of information potential granules.

A Comprehensive Analysis of the Impact of Selecting the Training Set Elements on the Correctness of Classification for Highly Variable Ecological Data

Classification of objects in empirical data, especially in biological sciences, is a very complex process and has been a big challenge for researchers who do not specialize in data analysis. Therefore, in this study, we present a comprehensive summary of selected classifiers operating on both exact and fuzzy numbers. The results of performance of specific classifiers are compared on the example of a unique set of empirical data on changes in the behavior of animals in response to environmental factors. This is one of the key challenges in ecological research and it is strictly related to ecosystem changes caused by climate change. Nowadays, changes in behavior are a very popular topic of research because as a result of the COVID-19 pandemic and lower activity of people (lockdown effect). Therefore, various unusual reactions of wild animals were found around the world. A detailed compilation of research results, shortcomings, and strengths of various classification methods may be a compendium of knowledge for biologists and other practitioners as well as researchers working with empirical data.