Star Crossed, Validated Love with Perfect Signs
Can astrology guarantee you a partnership that grants a lifetime of happiness? What about scientific, data-driven algorithms that match profiles in dreamy love connections? Three students at UC Berkeley set out to do just this by combining the ancient philosophies of zodiac compatibility with cutting-edge machine learning methods in their revolutionary product, Perfect Signs. Unsatisfied with the failure to provide insights into astrological romantic compatibility by apps such as Costar or Hinge, the trio built a solution that affords these answers. Their pilot website lets anyone take a short survey and get a breakdown of the zodiac sign and characteristics they should look for in their next romantic partner.
While based on the traditional twelve zodiac sun signs, Perfect Signs uses its own data on successful couples to make its recommendations based on historic patterns rather than theoretical matches as often found on websites like Café Astrology. The trio especially loves their Trends page that lets people take a peek behind the statistical curtain and learn more about popular characteristics and matches with their own sign – fun and informational!
This Python app built using Dash and runs Decision Tree Classification and Logistic Regressions through Scikit-learn on various responses given by a user to generate its results. The team envisions their algorithm being incorporated into existing dating apps in the market as a unique factor that differentiates these platforms from other competitors as well as an opportunity for more widespread data collection to refine and perfect their model. Perfect Signs is sleek in its simple concept and design, catering to audiences of all ages and versatile in its use for both research and entertainment. There’s no product quite like this one on the market! Try it now at perfectsigns.herokuapp.com.
Project by: Aastha Jha, Devina Sen, Varsha Sundar