Project Check-in #3 Reflection
Introduction: We plan to predict stocks using three different models (1D CNN, RNN, Transformer), and we will compare and contrast the results of each.
Challenges: The hardest part of programming so far was to figure out the corresponding function names and features between Tensorflow and PyTorch. Since we are implementing this project with PyTorch but we did all our code for this course using Keras, so PyTorch has been a new framework that we had to learn about in order to use.
Insights: Currently, we are getting results from the 1D CNN and Transformer implementations. However, we haven't decided on what timeframe to predict (for example, use the past 60 days of stocks to predict the next 30 days) so we need to do that before we can compare the performances.
Plan: We are a bit behind on our project because we need to complete the RNN implementation. We also need to dedicate a significant amount of time to the writeup and poster so we can print the poster on time.
Log in or sign up for Devpost to join the conversation.