Since the start the COVID-19 Lockdown, I've found myself exploring my interests in ways I never had before. Despite the isolation, I found new joys in the many new hobbies I picked up as many did many of my friends. I wanted to create a tool that would help people find new hobbies that are more likely to suit their personality and lifestyle.

What it does

Personality Neighbors takes a data-driven approach to help users to find new hobbies based on their innate characteristics, lifestyle, and habits. We recommend the top musical genres and hobbies of the happiest individuals in the "neighborhood". The ultimate goal is for users to find new ways to cope and find joy in the midst of the pandemic and de-stress from the work-from-home life.

How we built it

Personality Neighbors leverages a dataset of over 1,000 young people who answered a questionnaire about their interests as well as over 50 detailed personality traits. A nearest neighbors algorithm was implemented to find 10% of the surveyed population that is most similar to the user. Within that group, the tool looks at the happiest individuals based on their response to questions regarding energy and happiness. The tool then returns the top hobbies and musical genres that those individuals consume.

Challenges we ran into

The first challenge was finding a dataset that had both metrics of happiness and hobbies. Once a sizeable dataset was found, the next challenge was determining the best machine learning approach. There were many possibilities such as building a classifier to score different hobbies or utilizing the nearest neighborhood approach. Lastly, we could not ask users to answer over 50 personality questions on the spot, so a feature selection approach was employed to determine the features that play the biggest role in determining the happiness rating of an individual. A regression approach was used so that coefficients could be used to determine feature importance of the different personality traits.

Accomplishments that we're proud of

I am proud of implementing a user interface that quickly adapts to user input. In real time, users can see how their choices affect the rankings of recommendations.

What we learned

I learned how to produce an RShiny website and create recommendations that are relatively straightforward to understand.

What's next for Personality Neighbors

In the future, we would like to enable users to explore the characteristics of their "neighbors" beyond hobbies. In future versions, people can explore their neighbors' phobias, and the finer details of their personalities such as spending habits and work habits. We will also add word clouds that allow users to visualize the characteristics strengths of them and their "neighbors".

Personality Neighbors:


Data references:

Built With

  • kaggle
  • r
Share this project: