Project Background: Launched in 2004, OkCupid pioneered inclusivity and a unique question-based matchmaking approach. Its signature feature is a match percentage algorithm leveraging user responses. This project aims to refine this system by delving deeper into user data.
Objective:
- Identify distinct user clusters based on personality, values, and preferences.
- Uncover the key dimensions influencing compatibility within these clusters.
- Ultimately, enhance the accuracy and effectiveness of OkCupid's matchmaking process.
Data and Methodology:
- A dataset of 59,946 OkCupid users with 31 attributes was utilized.
- We leveraged the power of Word2vec, a superior technique compared to TF-IDF, to analyze user-generated text data and calculate profile similarity with greater accuracy (achieving 0.99% similarity compared to 0.21%).
- Machine learning algorithms were deployed to unlock deeper insights into user preferences, personalities, and compatibility factors.
Expected Outcomes:
- Identification of distinct user clusters with unique characteristics.
- Discovery of the key dimensions driving compatibility within each cluster.
- Development of a refined matchmaking algorithm using insights gleaned from user data analysis.
- Increased matching accuracy, leading to higher user satisfaction and success rates.
- Enhanced platform reputation and potential for attracting a larger user base.
Benefits: This project holds the potential to revolutionize OkCupid's matchmaking system, leading to a more personalized and effective user experience. By creating more compatible matches, user satisfaction and platform loyalty will likely increase. The project's findings can also inform future platform development and marketing strategies.