Loading...
Publication
Extração de conhecimento através de Data Mining: obtenção de regras de associação num dataset público
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 950.66 KB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
Este artigo apresenta um trabalho de investigação que consistiu na Análise de Carrinhos de Compras (Market Basket Analysis) através da aplicação de técnicas de data mining a um
conjunto de dados de dimensão significativa. Pretende-se encontrar conjuntos de itens habitualmente comprados em conjunto e daí gerar Regras de Associação, as quais podem ser valiosas para
campanhas promocionais ou sistemas de recomendação. Foi adotada a metodologia de investigação
CRISP, especialmente vocacionada para data mining, e são descritas as diversas fases da mesma,
com foco na análise exploratória dos dados (para os conhecer), a preparação dos dados para aplicação e configuração do algoritmo Apriori e o estudo do comportamento do modelo obtido, aumento
e diminuição do número de Regras de Associação geradas, de acordo com a variação dos parâmetros e dos dados utilizados.
This paper presents a research that consisted of a Market Basket Analysis through the application of data mining techniques to a large dataset. The aim is to find sets of items commonly purchased together and then generate Association Rules, which can be valuable for promotional campaigns or recommendation systems. The CRISP research methodology was adopted, which is especially designed for data mining research. Its various phases are described, focusing on the exploratory data analysis (to understand better the data), the preparation of data for application and configuration of the Apriori algorithm and the assessment of the model behaviour, measured by the increase and decrease of the number of generated Association Rules, according to the variation of the parameters and the selected data
This paper presents a research that consisted of a Market Basket Analysis through the application of data mining techniques to a large dataset. The aim is to find sets of items commonly purchased together and then generate Association Rules, which can be valuable for promotional campaigns or recommendation systems. The CRISP research methodology was adopted, which is especially designed for data mining research. Its various phases are described, focusing on the exploratory data analysis (to understand better the data), the preparation of data for application and configuration of the Apriori algorithm and the assessment of the model behaviour, measured by the increase and decrease of the number of generated Association Rules, according to the variation of the parameters and the selected data
Description
Trabalho apresentado em XXX Jornadas Luso-Espanholas de Gestão Científica, 5-8 fevereiro 2020, Bragança, Portugal
Keywords
Data Mining Regras de Associação Algoritmo Apriori Association Rules Apriori Algorithm