El Dorado 🏆
As with the instability brought by COVID-19, many people can't physically visit places to find the perfect city for them, if they are moving. Students are graduating and having to relocate for jobs, people are shifting closer to family and people are still having to move from places to places. There is a large number of people that need a city recommendation system upon their preferences, so they don't have to surf the internet for hours to find the perfect city, the code does it for them.
What it does
From the inputted preferences, it outputs the user a list of cities based on the initial recommendation. Also, it provides a one-stop-shop for your city searching needs.
How we built it
For the first couple of hours, we invested heavily in acquiring data: we collected seven to eight large data sheets for use. We leveraged stream lite technology for the rapid development of a visualization tool. Furthermore, we performed a large scale of data wrangling and data formatting and filtering. The data engineer tasks were performed in various environments, such as Kaggle notebooks, VS Code, and Jupyter notebook. Also, we leveraged zip to the city library to convert and connect different datasheets. We used markdown pages to create an about us page on the website. We performed clustering and unsupervised ranking algorithms. However, we were unsuccessful in our ability to use the trained model for our recommendation system.
- Coding Languages - Python, Jupyter Notebook, and bash
- Python Libararies - Seaborn, Matplotlib, Pandas, Numpy, ScikitLearn, Streamlit, and Zipgrabber.
Challenges we ran into
- Working remotely was a burdensome challenge for us because we could not coordinate as well as if we were working together in person. To keep up to date, we needed to communicate effectively via Discord and GitHub Project To-Do Lists.
- Since so many different individuals were working on the project simultaneously, it was a challenge for the dashboard to be consistent and not riddle with bugs. To speed up the prototyping process and brainstorming process, we used whiteboard and gimp to communicate ideas visually.
Accomplishments that we are proud of
- Overcoming the above challenges was our biggest accomplishment.
- Utilizing the knowledge we gained from the workshops to create our application.
- Working with a new tech stack and stepping out of our comfort zone
- Learning that this is a novel idea that others have not created before.
What we learned
We learned the intercourse of data searching, wrangling, and interactive dashboard development. We also learn how to intergrade data from different sources, and preparing the data. It was some of our first times working with big data and using streamlit to host it as a web platform so the learning curve was steep but we made the best effort to overcome those obstacles and glad to say we made it out alive, yay!
What's next for El Dorado
In the current website, we have a text box for users that will take in the input of phrases in plain English to describe their next city and comprehend it and output their desired city. In the future, we will be using this to gain an information data absorbing tool, to understand the link between words and association the cities. This dataset can be used in NLP to make our ranking algorithm, more robust and more rewarding for the use. Another thing we believe we can improve upon is the increase in the number of cities that are available in the search engine, we currently only have 20,000 cities that are only domestic. A big step forward for us would require us to expand our reach all around the world.