APPLICATION OF DATA MINING IN MARKETING USING E-COMMERCE CUSTOMER SENTIMENT ANALYSIS

In this study, sentiment analysis was developed and applied to technological products in the Twitter/X social network, also, the opinions expressed by customers were determined and finally the most suitable predictive model derived from Machine Learning was identified. For this purpose, 7102 tweets...

Descripción completa

Detalles Bibliográficos
Autores: PETRLIK, IVAN, Coveñas Lalupu , José, CARRANZA BARRENA, WILFREDO, TORRES TALAVERANO, LUZ
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2024
País:Perú
Institución:Universidad de San Martín de Porres
Repositorio:Revistas - Universidad de San Martín de Porres
Idioma:español
OAI Identifier:oai:revistas.usmp.edu.pe:article/2855
Acceso en línea:https://portalrevistas.aulavirtualusmp.pe/index.php/rc/article/view/2855
Access Level:acceso abierto
Palabra clave:Minería de datos, análisis de sentimientos, aprendizaje automático, e-commerce
Data mining, sentiment analysis, machine learning, e-commerce
Extração de dados, análise de sentimentos, aprendizagem automática, comércio eletrónico
Descripción
Sumario:In this study, sentiment analysis was developed and applied to technological products in the Twitter/X social network, also, the opinions expressed by customers were determined and finally the most suitable predictive model derived from Machine Learning was identified. For this purpose, 7102 tweets related to Apple and Samsung products were collected, using the methodology proposed by Erl, Khattak and Buhler which facilitated the implementation of its critical phases. The results obtained from sentiment analysis were evaluated using standard metrics such as Accuracy, Precision, Recall and F1-Score, applied to four machine learning models: K-Nearest Neighbors (KNN), Logistic Regression (LR), Random Forest (RF) and CatBoost Classifier (CC). Of these, the CatBoost Classifier proved to be the most effective, achieving 89% in Accuracy, 90% in Precision, 89% in Recall and 88% in F1-Score. It was concluded that the CatBoost Classifier model was the optimal model for analyzing sentiment on Twitter/X, due to its ability to provide valuable insights into the perception of promoted technology products enabling effectiveness in digital marketing campaigns.