Publicação

Customer Review Analysis

Ver documento

Detalhes bibliográficos
Resumo:Over the last years, Cerascreen has grown rapidly and expanded into more than 20 countries, always focusing on offering more diverse products, supplements, and services. Unfortunately, it collected a lot of data during these years, which was not yet stored, losing valuable insights. In a new initiative Cerascreen wants to be the most trusted digital predictive health platform. Therefore, it intends to utilize its data to understand its customers better and offer superior products and services according to the customer’s needs. The focus of this internship report was to find a way to analyze Cerascreen’s customers’ reviews to understand its customers better and respond to properly the given feedback. In addition, since the reviews have not been stored before, this report also deals with review retrieval. An exploratory data analysis of the reviews’ ratings and texts was conducted to find the first significant insights. The investigation found that although the overall review consensus was positive, it differed by country, while the reviews’ length was related to their ratings. A topic model was developed to find more information on what customers are talking about. The Model was able to find several different topics, including product-, supplement-, and servicespecific reviews. Lastly, a newly created key performance indicator about customers satisfaction uses the new insights about the ratings and the review topics, which a dashboard partially visualized through a dashboard.
Autores principais:Tueschen, Philipp
Assunto:BERTopic Sentence Embeddings Text Mining Topic Modeling Unsupervised Learning
Ano:2022
País:Portugal
Tipo de documento:dissertação de mestrado
Tipo de acesso:acesso aberto
Instituição associada:Universidade Nova de Lisboa
Idioma:inglês
Origem:Repositório Institucional da UNL
Descrição
Resumo:Over the last years, Cerascreen has grown rapidly and expanded into more than 20 countries, always focusing on offering more diverse products, supplements, and services. Unfortunately, it collected a lot of data during these years, which was not yet stored, losing valuable insights. In a new initiative Cerascreen wants to be the most trusted digital predictive health platform. Therefore, it intends to utilize its data to understand its customers better and offer superior products and services according to the customer’s needs. The focus of this internship report was to find a way to analyze Cerascreen’s customers’ reviews to understand its customers better and respond to properly the given feedback. In addition, since the reviews have not been stored before, this report also deals with review retrieval. An exploratory data analysis of the reviews’ ratings and texts was conducted to find the first significant insights. The investigation found that although the overall review consensus was positive, it differed by country, while the reviews’ length was related to their ratings. A topic model was developed to find more information on what customers are talking about. The Model was able to find several different topics, including product-, supplement-, and servicespecific reviews. Lastly, a newly created key performance indicator about customers satisfaction uses the new insights about the ratings and the review topics, which a dashboard partially visualized through a dashboard.