A stock price predictro using Regression and LSTM
This was developed as my undergrad project. Stock price movement analysis is one main study area in algorithm trading. Although nobody in this world can predict the next-moment stock prices with an absolute 100% accuracy, the stock price change pattern is still one of the main interests of many investors. This project was made keeping in mind the small, novice investors who can benifit from a non-biased application based recomendation on purchasing, selling or holding stocks.
- Python
- Machine Learning
- Linear Regrassion
- LSTM
Downnload the 3 .ipyb files and open them in Jupyter notebook and run one cell at a time.
First we had to clean the data and add various new parameters. Wherever there were null or invalid values we reaplaced this with the mean, this can be seen in the Reg-Clean&Prep-Process.ipynb file. We also clasify the day to day changes in few catogories namely Slight or No Change, Slight Positive, Slight Negative, Positive, Negative, Among Top gainers, Among Top Losers, Breakout Bull, and Breakout Bear. Below we can see the number of times the stock was in these catogories.
There are two files for the fitting and clasification for LSTM pls see this and for Regression pls look here.
-
Email - taher.mulla@gmail.com