Uncovering social media usage and interaction patterns using data analytics
Data Wow's client, a leading Japanese local social media app, is focused on creating a space where users can connect with others who share their interests and build strong relationships. The client faces challenges such as acquiring new users, retaining existing users, and continuously engaging them. Hence, Data Wow tasked SMU students to leverage data analytics and machine learning techniques to:
- Discover typical patterns of user networks and interactions through data discovery and exploration techniques.
- Recommend suitable circles (social network groups) to users who share common characteristics, to keep them engaged.
Data Wow provided the students with user demographics and three months’ interaction data from the social media network. To better understand the business and data, the students had regular communication with the project sponsor for clarifications. The students then prepared the data, translating it from Japanese to English, and used data exploration techniques to uncover hidden patterns within the data set. They experimented with different data visualizations to adequately present the uncovered insights. For the challenge of recommending suitable circles, students experimented with and evaluated various machine learning and recommendation models. Some of the important insights derived from the data analysis include that most users do not join circles, most users post outside of circles rather than in them, and most users are not interacting (e.g., not creating posts) within the network. This highlights the huge potential for applying machine learning models to recommend suitable circles and posts to users, enhancing their interaction and engagement within the network.
Through this project, the students gained valuable experience in both technical and business aspects. On the technical side, working on real-life projects allowed them to apply their classroom knowledge, such as various visualization techniques and methods for data cleansing and processing. On the business side, the students learned the importance of presenting data in a way that is easily understandable to non-technical stakeholders. Translating the initial data set and business challenges into final project artifacts required not only their technical skills but also their critical thinking and problem-solving abilities. Additionally, the students improved their collaboration with teammates and their ability to communicate ideas effectively to peers and industry partners.
"I was impressed by the enthusiasm and curiosity of the students in the SMU-X program. Their works show opportunities hidden beneath our Social Interaction datasets and help us understand how communities develop in our app." -- Supaseth Wongondee, Data Scientist, Data Wow
"The students from SMU-X have impressed us with their diligence and competence. They presented many valuable insights and solutions for building a recommendation system on our social networks. Their dedication and creativity have been truly remarkable." -- Thammasorn Harnpadungkij, Data Science Specialist, Data Wow
"Throughout this project experience, I discovered the importance of adapting to project timelines, particularly due to the non-linear process and unexpected data limitations unique to this setting, contrasting with structured academic projects. A notable challenge revolved around working within the confines of provided data while upholding client confidentiality. Meanwhile, a major success for me was successfully delivering a coherent project that met the client's requirements, following weeks of collaborative efforts guided by professors. Additionally, collaborating with an overseas team for a week brought distinct challenges, requiring constant adjustments following each consultation. Reflecting on the experience, I recognize the need to prioritize enjoying the project journey more and ensuring personal well-being, despite occasional concerns raised by team members about my rest and eating habits. This project provided a unique opportunity for international collaboration and hands-on experience in data consulting, setting a strong foundation for future endeavors, including the upcoming Analytics Applications module. With that, I would like to extend a heartfelt thank you to Prof Keith and Prof Graham for their mentorship throughout the project!" – Eleora Emma Jane Lim, Student in the IS455 course