Análise de consumo de produtos à base de leitelho através de técnicas de mineração de dados

Abstract

Buttermilk is a by-product of butter production, and its composition is similar to that of skim milk. Despite the high production of butter in Brazil and the nutritional and industrial values that buttermilk has, this co-product is not easily found as an ingredient, and its disposal is highly polluting when done incorrectly. Thus, we sought to compare the trends and consumption patterns of buttermilk-based products by people inside and outside Brazil. For this, we used the results of the form "Estudo do comportamento brasileiro sobre o consumo de produtos lácteos" and we also collected, with data scraping, details of products with buttermilk in online stores and their evaluations, along with basic information of the evaluators, profile picture, and country. The opinions obtained in the two data sources were categorized, with sentiment analysis of the textual content, by polarity (positive, negative, or neutral), and sex and age group were determined by processing the valid profile images collected in the data scraping. Among the respondents, 58% were not familiar with the term “buttermilk”, of which 40% answered that they would not consume products based on it because it would have “bad taste” or “bad texture”. “Presence of substances not suitable for human health” and “inferior nutritional quality” were also frequent responses, representing 39% of the total, despite other studies already showing the opposite. It was possible to observe that, despite the benefits of buttermilk, very few variety of industrialized products use it in the composition, little knowledge and also little consumption of buttermilk and its products were identified by the responses to the form. However, there were few negative answers for “word, feeling and/or emotion” about buttermilk, and the main reasons for trying buttermilk products are mainly flavor and curiosity. In addition, the categories of products that participants are most interested in trying, such as baked goods, are aligned with the categories of international products collected. The estimated age distribution for users who evaluated buttermilk-based products was similar to that of the participants in the form, mostly between 25 and 44 years old. The evaluations collected were also mostly classified with positive feelings, showing that there is a market to be explored, bringing benefits to the producer, to consumers, and also as a way to reduce possible inappropriate discards. Future studies should improve data scraping, such as increasing the frequency of scraping or collecting more information from users who evaluated purchases for purchase profile mapping.