Inspiration
We were inspired by the wealth of information provided by TAMU datathon and wanted to provide users and future competitors an easy way to access all of the available resources through a For You page. Machine learning algorithms and data science tools were used throughout this project - we hope that our learning process is able to be conveyed through this project and inspire future competitors!
What it does
The user will log onto the page and will immediately be greeted by big picture visualizations of the other competitors' majors, classification and specific interests. Then, they are able to see their recommended 'dream' team based on clustering using parameters such as experience, university and age bracket. LinkedIn internships (with detailed visualizations) and video links inspire the user to learn more about data science and provide 'big picture' information.
We also have built an NLP search engine that outputs workshops based on user-inputted text/keywords by checking the words' relation to specific workshop 'tags' based on a logistic regression model.
How I built it
Google Colab, Pandas, Matplotlib and Sklearn were used for the back-end/data analysis tasks. For the front end, we worked on Streamlit for a neat representation of our work. Clustering algorithms, NLP and logistic regression were primarily used.
Challenges I ran into
Since we're relatively new to working with datasets, reprocessing the sets took the most time to complete and ended up not being usable for workshop clustering. We also tried using an API to link external data from LinkedIn onto the personalized "For You Page". However, as our team consists of beginners, the focus on learning new tools and libraries hindered speed of completion. While plotting the graphs, we also ran into a problem with assigning the correct colors on the legend. There was no way to specifically choose a particular color on plt.legend(), which may lead to some inaccurate data. Additionally, we had some issues with data representation on streamlit, but were able to learn a lot from the challenges we faced.
Accomplishments that I'm proud of
It seemed daunting at first to finish such a complicated task in around 24 hours, but we were able to create a functional page and deploy machine learning algorithms to accomplish a good user experience. The learning process was difficult but yielded great results.
What We learned
How to process datasets for ML analysis, deploying NLP, logistic regression, clustering, and visualization of complex datasets. The workshops throughout the day definitely broadened our knowledge on a variety of subjects related to machine learning and programming. This was our team's first datathon/hackathon, so it was an enriching learning experience!
What's next for TAMU: For You
Create clustering based on workshops for boot camp prediction, badge and achievement dashboards and an NLP to access external resources like articles or tutorials could add to the efficacy of the website.
Log in or sign up for Devpost to join the conversation.