The paper sheds light on the use of a self-learning GNG neural network for identification and exploration of the purchasing behaviour patterns. The test has been conducted on the data collected from consumers aged 60 years and over, with regard to three product purchases. The primary data used to explore the purchasing behaviour patterns was collected during a survey carried out among the elderly students at the Universities of Third Age in Slovenia, the Czech Republic and Poland, in the years 2017–2018. Finally, a total of six different types of purchasing patterns have been identified, namely the ‘thoughtful decision’, the ‘sensitive to recommendation’, the ‘beneficiary, the ‘short thoughtful decision’, the ‘habitual decision’ and ‘multiple’ patterns. The most significant differences in the purchasing patterns of the three national samples have been identified with regard to the process of purchasing a smartphone, while the most repetitive patterns have been identified with regard to the purchasing of a new product. The results significantly support the GNG network’s validity for identification of consumer behaviour patterns. The application of this method allowed quick and effective to identify and segment consumers groups as well as facilitated the mapping of the differences among these groups and to compare the consumption behaviour expressed by consumers on different markets. The identified consumer purchase patterns may play a basic role for marketers to understand consumer behaviour and then propose tailored strategies in international marketing.

The authors of the study put forward a hypothesis that it is possible to extend the forecast period for the models of discriminant analysis used to assess the risk of enterprise bankruptcy, focusing on the components of these functions in the form of one-dimensional predictors, i.e. the indicators most frequently included in the discriminant functions developed in Poland. Early warning about the growing risk of bankruptcy would be very valuable for any company. The dataset was constructed from all enterprises in Poland that went bankrupt in the years 2007-2013, which was the end of the research project period. Out of the 4,750 business entities that went bankrupt at that time, 2,739 filed financial statements with commercial courts. The main objective was realized using dynamic assessment of the variability of selected one-dimensional predictors of bankruptcy for all of these enterprises. Assessment of the time variability of the indicators under analysis allows conclusions on the predictive possibilities associated with early warning against insolvency of business entities. The results constitute input to the discussion on determination of the longest prognostic horizon that can be adopted in the models of discriminant analysis used to assess the risk of enterprise bankruptcy. Most of them cover an annual forecasting horizon. Only a few authors have attempted to construct models based on data from the two, three, or even four years preceding bankruptcy. The study showed that the main symptoms of the growing risk of bankruptcy in most of the surveyed enterprises are visible much earlier than one year before bankruptcy. This provides an opportunity to correct the predictive models and more time to restructure the company, to prevent bankruptcy. Therefore, the authors of the study have assessed the possibility of extending this forecast period.

The twenty-first century has been a time of significant changes in the lives of many societies worldwide. The constantly increasing access to the web has changed many aspects of social life, including the availability of information and opinions expressed by hundreds of millions of people around the globe. Social networking sites have become the most important forum for exchanging views on virtually any topic. Twitter has a dominant position here, transmitting millions of posts from their users every day. The researchers analysing the content of the data derived from social networks often experience difficulties in identifying the age of the published texts’ authors. Such information, with accuracy to specific age groups, can be very useful in e.g., marketing research, which provides knowledge on purchasing preference differences. In many cases, users do not publish any information about their age, which significantly limits user profiling possibilities, for instance. Linguistic research shows that representatives of different age groups use distinct vocabulary and grammatical forms. It seems that they are also distinguished by the use of different graphical symbols such as emoticons, emojis, pictograms, and other graphics used in text. The research presented aims to distinguish Twitter user age groups in distribution by the typical emojis used in the messages published. An attempt will be made to answer the question of whether the type and the number of the graphical symbols used in tweets can indicate the age group of their authors. Text mining methods and artificial intelligence were used to carry out the research. Keywords Twitter Text mining Cluster analysis Emoji Emoticon Social media

Textual data (textposts) account for a significant portion of all data posted on the Internet. One piece of information that researchers are seeking to obtain about the authors of textposts is their age, which is not always made public, yet important from the point of view of marketing, social and economic research. Language research shows that representatives of different age groups tend to use a distinct set of vocabulary and grammatical forms. Presumably, textpost formatting as well as the level of the correctness of the text itself may also differentiate user age groups. The aim of the research presented in this article is to use the elements typically eliminated from texts during text mining processes, such as emoticons, punctuation marks and words that are not content carriers (stopwords) to distinguish the age groups of the authors of Twitter (currently X) posts. The study analysed nearly 3 million tweets in English posted before July 2020. The research shows that distinguished textpost elements differentiate the age groups only to a small extent. The youngest users stood out the most due to their specific language characteristics in textposts.