Personally

Predict the psychological traits of a person on the basis of his/her social media interactions.

Inspiration

Our product is targeted towards people who want to gain insights about other people from their social media interactions, for example companies and universities can use it to find out about the personality traits of their candidates. Check out the colab notebook (link in github readme) to see how we trained the algorithm.

Here's How it Works:

text2

What it does

This project aims to predict the psychological traits of a person on the basis of his/her social media posts. It predicts the following traits using a random forest algorithm that we trained.

  1. EXT - Extraversion
  2. OPN - Openness
  3. NEU - Neuorticism
  4. CON - Conscientiousness
  5. AGR - Agreeableness

How we built it

  • Firstly we tried to scrape the user's social activities based on their usernames or some keywords.
  • Passed this into a model and getting the results of the user's personality
  • Integrating with a web app
  • Creating an awesome UI so that users online can get an insight into their social activity by just some keywords

Props!

  1. Python
  2. Socialreaper
  3. SkLearn
  4. matplotlib
  5. Pandas
  6. GSearch Library

Challenges we ran into

Scraping from sites was tough and correctly processing this information as well. Furthermore, we had to make sure the results were produced fast as that was important for a good user experience. Making UI look good but classy was also important and it took time, but we did it!

Accomplishments that we're proud of

  • A project that could have many use cases such as using it to prevent cyber-bullying, fraud
  • Analyzing something cool like similarities between genius people.

What we learned

  • We learned how to scrape social media posts.
  • Learned how to efficiently use regex to remove urls and special characters from text.
  • Integrating sklearn models with a web app
  • Building projects in short timeframes :)

What's next for Personally

  • Expanding the project with some more of the use-cases
  • We could also use it for sites such as LinkedIn which could help users to analyze what they might be doing wrong if they are not getting a job and seeing an analysis of other people to correct their mistakes

Built With

Share this project:

Updates