stock_prediction_LSTM

Inspiration

Successful implementation of stock market predictions could make significant progress for any individual or overall country's economic progression. Personally, I was motivated to do this project cause my keen interest in investment market. In addition to that, experience with LSTM and forecasting time series is interesting to learn.

What it does

Essentially, it takes data from year 2010 to 2016 to make stock price predictions for subsequent time-frame followed. Here the machine learning algorithm I choose after some experiments with moving average, ARIMA is LSTM. LSTM chooses what to forget and what to insert into memory, allow a network to choose a path to focus on in the visual field

How I built it

I started the project with choosing appropriate dataset from kaggle I then utilized python's packages for data wrangling and EDA Studied some research papers and online resources for LSTM, statisitcs and stock price

Challenges I ran into

Initially, I decided to implement this projects using microsoft AzureML and data science virtual machine However, after some rendering here and there I couldn't manage to run it without exhausting memory usage I switched to jupyter notebook and keras afterwards

Accomplishments that I'm proud of

Learning more about LSTM and successfully implementing on selected dataset

What I learned

Understanding why would one need to be able to predict stock price movements Predicting and visualizing future stock market with current data and what potential impacts it could have

What's next for stock_prediction_LSTM ?

I could use most recent or live data to make this project more empirical and relevant

Built With

• keras
• lstm
• matplotlib
• numpy
• pandas
• python
• scikit-learn