Back to Projects

Understanding JKN Mobile User Experiences: A Sentiment Analysis and Topic Modeling Approach

October 2024
Sentiment Analysis
Topic Modeling
JKN Mobile App
User Experience
Understanding JKN Mobile User Experiences: A Sentiment Analysis and Topic Modeling Approach

Overview

Analyzed 27,000+ JKN Mobile user feedback, identifying key sentiments and concerns using Python and NLP (LDA), uncovering 3 key user concerns, and driving actionable product enhancements.

Key Achievements

  • 3 Key User Concerns Identified: Discovered the most pressing issues affecting user experience
  • Actionable Insights: Provided specific recommendations that led to product improvements

Background

The JKN Mobile app is a digital platform designed to facilitate access to National Health Insurance (BPJS Kesehatan) services. It offers various features, including checking insurance status, finding healthcare providers, and managing claims.

Despite the JKN Mobile app's 4.3 rating on the Play Store, there are anecdotal reports of user dissatisfaction and challenges, such as difficulties in creating accounts and receiving OTP codes. This project seeks to systematically investigate these issues and identify areas for improvement.

Methodology

  1. Data Collection: Scraping user reviews from Google Playstore using 'google-play-scraper' library.
  2. Data Cleaning: Remove duplicate reviews, handle missing value.
  3. Text Cleansing: Remove punctuations, emoji, stopwords, handle slang words, ensuring the text is clean to proceed further.
  4. Exploratory Data Analysis: The data was thoroughly examined to understand its characteristics and identify potential patterns.
  5. Sentiment Analysis Labeling: Labeling the reviews using the Indonesia-bert-sentiment-classification model available in Hugging Face.
  6. Text Analysis: Drawing bigrams, trigrams, word clouds from each sentiment category to unleash pattern.
  7. Topic Modeling: Employing LDA to identify the main themes and topics discussed in the reviews.
  8. Dashboard creation: For interactive visualization and easy customization, Streamlit was used to create dashboard.
  9. Create narrative report: Present the findings and comprehensive analysis in a narrative format, published in medium.

Results & Insights

Topic Distribution

The most frequently discussed topic among JKN Mobile users was the difficulty of registering and logging in due to OTP code errors. This issue was prevalent in negative reviews, indicating a poor user experience. Bigrams and trigrams also supported this finding, with users mentioning the challenges of waiting for OTP codes and attempting multiple registrations without success.

Neutral and negative sentiments were evenly balanced, with 37.5% of users expressing neutral feelings and 37.4% expressing negative sentiments. The word cloud analysis of neutral sentiment revealed that users often discussed registration, login, OTP codes, and verification processes, suggesting a need for assistance or improvements in these areas.

Users expressed frustration with three main functionalities:

  • Registration and login: Difficulties with OTP codes, long wait times, and multiple failed attempts were common complaints.
  • Captcha errors: Some users encountered issues with the captcha code during registration and login.
  • Health facility registration: Users found it challenging to register for a health facility.

On the positive side, users appreciated the app’s ability to access JKN services and its ease of switching health services. The following areas require improvement:

  • Registration and login function: Investigate the reasons for difficulties with OTP codes, captcha errors, and multiple failed attempts and implement solutions to address these issues.
  • Online registration feature: Investigate the reason for the difficulties in registering for a health facility.

Recommendations

To enhance user satisfaction and address the identified issues, the following recommendations are proposed:

  • Streamline the registration and login process: Simplify the verification process, reduce wait times for OTP codes, and implement more robust error-handling mechanisms.
  • Improve OTP delivery reliability: Explore alternative methods for OTP delivery, such as email, or in-app notifications, to ensure timely and reliable receipt.
  • Enhance captcha functionality: Implement a more user-friendly and accurate captcha system to reduce errors and improve the login experience.
  • Provide clear instructions and guidance: Offer detailed instructions and guidance on the registration and login process, including troubleshooting tips and FAQs.
  • Consider implementing biometric authentication: Explore the use of biometric features (e.g., fingerprint, facial recognition) as an additional or alternative authentication method to improve security and convenience.

Categories

machine deep-learning
data analysis

Objectives

  • Analyze user sentiments towards the JKN Mobile application
  • Identify key topics and concerns from user feedback
  • Provide actionable insights for application improvement

Tools & Technologies

Python
NLTK
Scikit-learn
LDA
Pandas

Data Source

27,000+ user reviews from Google Play Store and App Store