Sentiment Analysis on Natural Skincare Products

Abstract

Skincare Industry was increasing rapidly year by year. In contrast, many skincare companies have
brought their products and originality to attract many customers. However, due to many
controversial cases involving the chemical substance in skincare products, the company switched
to something more natural: natural skincare. With much natural skincare in the shop, many
customers face the problem of which one to buy. This research helps customers by giving a
guideline for the customers to make the decision. Sentiment Analysis is used to analyze the reviews
from past customers and create a visualization containing positivity and negativity of all the
reviews. Five classifiers were used to produce the best result: Naïve Bayes, KNN, SVM, Decision
Tree, and Deep Learning. The reviews were collected from Sephora.com websites, and the tools
used in analyzing the reviews are Python and RapidMiner. Reviews collected are 10000 data from
a website. The result shows that Deep Learning and Decision Tree are classifiers in sentiment
analysis with almost 80% accuracy and 60% F1 measurement. F1 measure is a measure of a test's
accuracy. For future enhancements, the data collected can be more than this research, and no data
imbalance was created.