Sentiment analysis on footwear products preferences based on Twitter feeds

Consumer's preference on footwear products refers to the consumer's choices and evaluations on footwear's attributes before they purchase the product. This study focuses on analysis of consumer preferences for footwear products based on Twitter data. With increasing predicted revenue...

Full description

Bibliographic Details
Published in:AIP Conference Proceedings
Main Author: Azman M.L.; Rosdin N.Y.N.; Jasni N.H.; Nasir N.
Format: Conference paper
Language:English
Published: American Institute of Physics 2024
Online Access:https://www.scopus.com/inward/record.uri?eid=2-s2.0-85203120159&doi=10.1063%2f5.0224032&partnerID=40&md5=e52f47e3eec9f06619df6fb665f3eade
id 2-s2.0-85203120159
spelling 2-s2.0-85203120159
Azman M.L.; Rosdin N.Y.N.; Jasni N.H.; Nasir N.
Sentiment analysis on footwear products preferences based on Twitter feeds
2024
AIP Conference Proceedings
3123
1
10.1063/5.0224032
https://www.scopus.com/inward/record.uri?eid=2-s2.0-85203120159&doi=10.1063%2f5.0224032&partnerID=40&md5=e52f47e3eec9f06619df6fb665f3eade
Consumer's preference on footwear products refers to the consumer's choices and evaluations on footwear's attributes before they purchase the product. This study focuses on analysis of consumer preferences for footwear products based on Twitter data. With increasing predicted revenue in 2023, the worldwide footwear consumer market is expanding, with factors such as brand and design influencing purchase decisions. The lack of information in understanding consumer preferences prevents footwear companies and related groups from receiving useful insights to enhance or develop products. The study aims to identify topics and sentiments related to footwear products on social media platform, Twitter, through text mining to benefit society, social media publicity, footwear companies and related groups in understanding consumer preferences. RapidMiner software is utilized in this study. The method includes data retrieval through web scraping, text pre-processing, sentiment analysis, and topic modelling. The bag-of-n-grams model and Latent Dirichlet Allocation (LDA) are used to identify keywords and frequent topics in tweets, while sentiment analysis is conducted using Naïve Bayesian, Decision Tree, and k-NN classification models. The study also utilizes Independent Sample T Test to test hypotheses regarding sentiment score differences between different footwear brands mentioned in the tweets. From the analysis it is shown that the most frequent keywords such as "shoe", "nike", "boot", etc. and ten topics are identified, including "Shoe Brands and Styles"and "Women's Running Shoes."According to the sentiment analysis, 44.05% of tweets are positive, 21% are negative, and 34.94% are neutral. Meanwhile, k-NN obtained the highest accuracy value for the classification model at 75.51% among other models. By examining Twitter trends, user comments and keywords that customers used to describe footwear products on Twitter, the footwear business can gain valuable insights that will help them improve current products or create new ones. The findings from this study may serve as a reference for future qualitative research on footwear brands based on topic modelling and sentiment analysis. © 2024 Author(s).
American Institute of Physics
0094243X
English
Conference paper

author Azman M.L.; Rosdin N.Y.N.; Jasni N.H.; Nasir N.
spellingShingle Azman M.L.; Rosdin N.Y.N.; Jasni N.H.; Nasir N.
Sentiment analysis on footwear products preferences based on Twitter feeds
author_facet Azman M.L.; Rosdin N.Y.N.; Jasni N.H.; Nasir N.
author_sort Azman M.L.; Rosdin N.Y.N.; Jasni N.H.; Nasir N.
title Sentiment analysis on footwear products preferences based on Twitter feeds
title_short Sentiment analysis on footwear products preferences based on Twitter feeds
title_full Sentiment analysis on footwear products preferences based on Twitter feeds
title_fullStr Sentiment analysis on footwear products preferences based on Twitter feeds
title_full_unstemmed Sentiment analysis on footwear products preferences based on Twitter feeds
title_sort Sentiment analysis on footwear products preferences based on Twitter feeds
publishDate 2024
container_title AIP Conference Proceedings
container_volume 3123
container_issue 1
doi_str_mv 10.1063/5.0224032
url https://www.scopus.com/inward/record.uri?eid=2-s2.0-85203120159&doi=10.1063%2f5.0224032&partnerID=40&md5=e52f47e3eec9f06619df6fb665f3eade
description Consumer's preference on footwear products refers to the consumer's choices and evaluations on footwear's attributes before they purchase the product. This study focuses on analysis of consumer preferences for footwear products based on Twitter data. With increasing predicted revenue in 2023, the worldwide footwear consumer market is expanding, with factors such as brand and design influencing purchase decisions. The lack of information in understanding consumer preferences prevents footwear companies and related groups from receiving useful insights to enhance or develop products. The study aims to identify topics and sentiments related to footwear products on social media platform, Twitter, through text mining to benefit society, social media publicity, footwear companies and related groups in understanding consumer preferences. RapidMiner software is utilized in this study. The method includes data retrieval through web scraping, text pre-processing, sentiment analysis, and topic modelling. The bag-of-n-grams model and Latent Dirichlet Allocation (LDA) are used to identify keywords and frequent topics in tweets, while sentiment analysis is conducted using Naïve Bayesian, Decision Tree, and k-NN classification models. The study also utilizes Independent Sample T Test to test hypotheses regarding sentiment score differences between different footwear brands mentioned in the tweets. From the analysis it is shown that the most frequent keywords such as "shoe", "nike", "boot", etc. and ten topics are identified, including "Shoe Brands and Styles"and "Women's Running Shoes."According to the sentiment analysis, 44.05% of tweets are positive, 21% are negative, and 34.94% are neutral. Meanwhile, k-NN obtained the highest accuracy value for the classification model at 75.51% among other models. By examining Twitter trends, user comments and keywords that customers used to describe footwear products on Twitter, the footwear business can gain valuable insights that will help them improve current products or create new ones. The findings from this study may serve as a reference for future qualitative research on footwear brands based on topic modelling and sentiment analysis. © 2024 Author(s).
publisher American Institute of Physics
issn 0094243X
language English
format Conference paper
accesstype
record_format scopus
collection Scopus
_version_ 1812871793804836864