Project 3 : Stocks-Sentiment-Analysis-NLP-LSTM
Stocks-Sentiment-Analysis-NLP-LSTM
Stocks Sentiment Analysis NLP LSTM
1. Overview
Goal
- Humungous data on social media about a stock which is impossible to go through and make decisions.
- Leverage technology to gain interesting insights and make better investment decisions.
Impact
- Based on the signals of the model investors can make confident and well informed decisions.
- Time saved by 95% by leveraging technology.
- Avoid decision fatigue and analysis paralysis.
Challenges Faced
- Collection of data from twitter through twitter API.
- Structuring and making sense of unstructured and noisy text data.
- Data cleaning and preprocessing.
Interesting findings
- For positive class the words : long, bullish, buy call, high, hold are very common.
- For negative class short, bearish, buy put, low very common.
- Both the classes have some common words like volume which can imply both sentiments and add noise in the data.