Inspiration
Understanding stock markets looked like a very interesting problem to tackle. Sentiment Analysis to predict changes in stock prices seemed both challenging and practical.
What it does
Predict the change in stock price using sentiment analysis
How we built it
We used the Keras library to train an LSTM architecture to classify text
Challenges we ran into
One big challenge was finding a good dataset. We could only get our hands on one with only around two thousand training examples. Also, the training data was unbalanced, skewing our predictions to one category.
Accomplishments that we're proud of
The model was able to predict if a certain headline would impact the stock prices in a positive, neutral or negative way.
What we learned
- To scrape news articles on the internet
- To use LSTM models for text classification
What's next for Stock price trend Prediction
Take social media into account, and predict numerical prices instead of relative change.
Built With
- beautiful-soup
- colab
- keras
- lstm
- nltk
- python
Log in or sign up for Devpost to join the conversation.