Penerapan Metode Weighted Sum Model (WSM) dan Klasifikasi Komentar pada Sistem Pendukung Keputusan Aplikasi MyPertamina Menggunakan Metode Random Forest

Abstract

The development of a comment application system from Google Play Store users as a source of information for
application developers in evaluating and improving applications from the findings of weaknesses or deficiencies.
One such application is the MyPertamina application as a digital platform used by the public in making transactions
to buy fuel or other services from Pertamina. The purpose of this study is to build a Decision Support System
(DSS) to be able to classify application user comments by maximizing the Random Forest algorithm, and provide
alternative assessments by applying the Weighted Sum Model (WSM) method based on certain criteria. The first
stage collects various comments from MyPertamina application users, then the second stage carries out text preprocessing namely normalization, tokenization, stopword removal, and stemming. The third stage classifies into
three sentiment categories, namely positive, neutral, and negative using the Random Forest algorithm. After the
classification results are known, the fourth stage continues, namely applying the WSM method to assess or create
a priority scale as an alternative decision, for example starting with improving features in the application or user
areas that are most affected or impacted based on weighted criteria determined from the number of negative
comments, satisfaction levels and urgency of the issue. The testing conducted with the Random Forest
2
classification model yielded an accuracy value of 86%. Furthermore, the dashboard visualization showed that the
inaccurate data category had the highest average value of 0.11. The WSM method was shown to be more effective
in providing recommendations for prioritizing decision-making in a systematic and measurable manner. The
development of this system is expected to help MyPertamina application developers evaluate user feedback more
efficiently and objectively. The usefulness of this research for the company is that the company can understand
user perceptions and will continue to improve service quality, which impacts user satisfaction. Based on the results
of this research, the system can classify comments automatically by implementing the Random Forest Algorithm,
which is capable of providing good performance with an accuracy of > 80%. Management can use this system as
a basis for decision-making for the development of the MyPertamina application, and application developers can
better understand user perceptions automatically and make strategic decisions by processing user comment
databases.