Intervalos de confiança percentis por reamostragem Bootstrap para dados de Temporal Dominance of Sensations
Arquivos
Data
2023-02-27
Autores
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Federal de Viçosa
Resumo
Temporal Dominance of Sensations (TDS) é um método de análise sensorial, em que se avalia a percepção dos atributos (sensações) durante um determinado período de tempo pré-fixado, em que o avaliador fica exposto ao produto. Essa avaliação consiste em atribuir 1 (um) para a sensação sentida e automaticamente 0 (zero) para as sensações não sentidas durante o tempo de avaliação. No presente estudo foram avaliadas cinco bebidas de uva com diferentes proporções de água e açúcar, chamadas de A, B, C, D e E. Com o objetivo de conhecer o atributo dominante entre “ácido”, “adstringente”, “aroma”, “doce” e “sabor”. Gerou-se bancos de dados com 46 segundos (0 – 45) para 16 avaliadores com 3 repetições para as 5 sensações pré-definidas mais uma coluna com “nenhuma sensação”. Foram geradas curvas para cada atributo por meio da proporção em que os avaliadores atribuíram o valor 1 para a sensação sentida em cada instante de tempo e plotadas em um mesmo gráfico. Com essas curvas determinou-se quais sensações e em quais instantes atingiram a proporção mínima para ser considerada uma sensação dominante. De posse dos dados coletados para a TDS, aplicou-se o teste de Qui- quadrado para aderência e avaliou-se em quais instantes de tempo, pelo menos uma sensação, foi considerada diferente das demais. Neste trabalho, optou-se por apresenta somente os resultados para os instantes de tempo de 13 segundos para a bebida A e de 10 segundos para a bebida C, com a finalidade de apresentar o método que é válido para os demais tempos e bebidas. Aplicou-se o método de reamostragem Bootstrap em cada linha para construir intervalos de confiança percentis e a distribuição empírica dos dados. De posse dos intervalos de confiança, pôde-se inferir sobre quais sensações diferem estatisticamente, ou não, das demais e quais foram consideradas dominante pela TDS. Os resultados obtidos com os intervalos de confiança via Bootstrap, estão em acordo com as curvas construídas nas análises TDS empregadas atualmente na área de ciência e tecnologia de alimentos, que são apenas descritivas, portanto, a proposta apresentada é uma alternativa de análise inferencial para estudos com TDS. Palavras-chave: análise sensorial; suco de uva; qui-quadrado; bootstrap; temporal dominance of sensations
Temporal Dominance of Sensations (TDS) is a method of sensory analysis, in which the perception of attributes (sensations) is evaluated during a certain pre-fixed period of time, in which the evaluator is exposed to the product. This evaluation consists of assigning 1 (one) to the sensation felt and automatically 0 (zero) to the sensations not felt during the evaluation time. In the present study, five grape beverages with different proportions of water and sugar, called A, B, C, D and E, were evaluated. In order to know the dominant attribute between "acidic", "astringent", "aroma", "sweet" and "taste". Databases with 46 seconds (0 – 45) were generated for 16 evaluators with 3 repetitions for the 5 predefined sensations plus a column with "no sensation". Curves were generated for each attribute by means of the proportion in which the evaluators attributed the value 1 to the sensation felt at each instant of time and plotted on the same graph. With these curves, it was determined which sensations and at which moments reached the minimum proportion to be considered a dominant sensation. With the data collected for the TDS, the chi-square test for adherence was applied and it was evaluated at which time instants, at least one sensation, was considered different from the others. In this study, we chose to present only the results for the time instants of 13 seconds for drink A and 10 seconds for drink C, in order to present the method that is valid for the other times and drinks. The Bootstrap resampling method was applied to each row to construct percentile confidence intervals and the empirical distribution of the data. With confidence intervals, it was possible to infer which sensations differed statistically or not from the others and which were considered dominant by the TDS. The results obtained with the confidence intervals via Bootstrap are in agreement with the curves constructed in the TDS analyses currently employed in the area of food science and technology, which are only descriptive, therefore, the proposal presented is an alternative of inferential analysis for studies with TDS. Keywords: sensory analysis; grape juice; chi-square; bootstrap; temporal dominance of sensations
Temporal Dominance of Sensations (TDS) is a method of sensory analysis, in which the perception of attributes (sensations) is evaluated during a certain pre-fixed period of time, in which the evaluator is exposed to the product. This evaluation consists of assigning 1 (one) to the sensation felt and automatically 0 (zero) to the sensations not felt during the evaluation time. In the present study, five grape beverages with different proportions of water and sugar, called A, B, C, D and E, were evaluated. In order to know the dominant attribute between "acidic", "astringent", "aroma", "sweet" and "taste". Databases with 46 seconds (0 – 45) were generated for 16 evaluators with 3 repetitions for the 5 predefined sensations plus a column with "no sensation". Curves were generated for each attribute by means of the proportion in which the evaluators attributed the value 1 to the sensation felt at each instant of time and plotted on the same graph. With these curves, it was determined which sensations and at which moments reached the minimum proportion to be considered a dominant sensation. With the data collected for the TDS, the chi-square test for adherence was applied and it was evaluated at which time instants, at least one sensation, was considered different from the others. In this study, we chose to present only the results for the time instants of 13 seconds for drink A and 10 seconds for drink C, in order to present the method that is valid for the other times and drinks. The Bootstrap resampling method was applied to each row to construct percentile confidence intervals and the empirical distribution of the data. With confidence intervals, it was possible to infer which sensations differed statistically or not from the others and which were considered dominant by the TDS. The results obtained with the confidence intervals via Bootstrap are in agreement with the curves constructed in the TDS analyses currently employed in the area of food science and technology, which are only descriptive, therefore, the proposal presented is an alternative of inferential analysis for studies with TDS. Keywords: sensory analysis; grape juice; chi-square; bootstrap; temporal dominance of sensations
Descrição
Palavras-chave
Análise de séries temporais, Amostragem (Estatística), Suco de uva - Avaliação sensorial - Métodos estatísticos., Teste qui-quadrado
Citação
TEMÓTEO, Alex da Silva. Intervalos de confiança percentis por reamostragem Bootstrap para dados de Temporal Dominance of Sensations. 2023. 52 f. Tese (Doutorado em Estatística Aplicada e Biometria) - Universidade Federal de Viçosa, Viçosa. 2023.