Inspiration
What it does
When user inputs a catchy news title, the Currentcy web application will output the predictions of several exchange rates (US to Euro, Japanese Yen, Canadian Dollar, etc) based on the news title.
How we built it
For frontend, we used Bootstrap (along with HTML and CSS). For backend, we used Python (Pytorch) to develop recurrent neural network model, that is, the LSTM. To obtain training data, we got data from both the XE.com API and a Kaggle dataset on the newspaper title happening on different days. We used Flask to integrate the backend and frontend code.
Challenges we ran into
Insufficient data, limited permission on API requests, inefficient training process that did not reduce the cost function.
Accomplishments that we're proud of
We managed to conquer the challenges aforementioned and produce a webpage that produces a sensible prediction!
What we learned
Machine learning fundamentals, data engineering, Flask.
What's next for Currentcy
Ability to accept more variety of user input.
Log in or sign up for Devpost to join the conversation.