Sincerely, Your Penpal
Pen pal matching & communication app (we will be matching users to their pen pal based on the interests they indicated, and use ML to determine the personality of the user based on their Twitter tweets). Once matched, pen pals have the ability to create customized letters and upload videos to send each other (with a delay in time, as based on their real-life locations)
Covid has changed the way how people interact with each other. In the middle of the pandemic, with no chance of attending events physically and meeting people with similar interests and ways of self-care, we thought it would be great to re-introduce the concept of letter-writing for meeting new people
What it does
With a machine Learning model using a Kaggle dataset to predict the personality type of a person. Data is taken from the past 50 posts of the users twitter account to get an accurate idea of the personality type. Additionally, people enter their interests (finance/economics/UI design/ programming/software development, etc) and ways of self-care (art therapy, mindfulness, yoga, running, gymming, eating healthy, etc). Based on these three parameters, we find your perfect match.
How we built it
- The Machine learning model was built using the scikit learn library available in python. The model was then stored using joblib. Flask is used in the backend to deploy the machine learning model on the website and collect data from the user.
- Figma was used to sketch the look and feel of the website. Posthaste, work was started on the front-end web development in HTML, CSS, JS.
Challenges we ran into
- Teammates located in different timezones, scheduling meets and passing on work took good documentation
- Getting good accuracy in the Machine Learning model, since the dataset was rather unbalanced.
- We had issues hosting our front end with Flask, because we lacked experience with this area. But we did learn the basics of it.
- Attempts was made to use the Figma for HTML plugin - but that gave very bad results. So the website was coded from scratch.
Accomplishments that we're proud of
- The clean UI design of our website
- Having a working Machine Learning model
What we learned
- Time management and regular check-ins to ensure project is on track
- Work division by pairing off people to ensure accountability partners
- Working across timezones and handing off work with proper documentation
- Learning various softwares/languages/concepts from each other - Figma, Flask, HTML, CSS, API
What's next for Sincerely, Your Penpal
- Improving the accuracy of the machine learning model by using deep learning models
- Deploying the website and making an mobile app
- Marketing campaigns on social media to build a support group of interested people
- Wellness campaigns so penpals with the same self-care methods can attend events virtually together
- Option for job/internship search and mentorship based on interests and career directions