Skip to content

Project to predict stock prices with Recurrent Neural Network in TensorFlow with client input as web application (with Flask).

Notifications You must be signed in to change notification settings

banluong/stock-prediction-rnn-web-app

Repository files navigation

stock-prediction-rnn-web-app

Disclaimer: I'm not a financial expert. This project is for educational and entertainment purposes only to demonstrate the application of TensorFlow/Keras, Dash and other visualisations.

Please consult a professional financial consultant for investing. Invest at your own risk.

About

Project to predict stock prices with Recurrent Neural Network in TensorFlow/Keras, fbprophet (library by Facebook for forecasting time series data) with client input as web application (in Dash) for forecasting.

Feedback much appreciated and I hope you have fun running these dashboards! :)

Image

Image of Tesla (TSLA) stock price over time with forecast, lower and higher intervals.

Dependencies

Install the following libraries before running stock_dash.py and stock_forecast.py.

pandas
dash
fbprophet
yfinance
plotly
datetime

Dash is the main library to create the dashboard application, while yfinance and fbprophet are the dependencies to get historic stock prices and forecasting. Plotly needed for visualisations on the dashboard and pandas for storing and reading the price data. datetime for yfinance to get values up to 5 years from end date (end date default is latest trading date.)

Updates and Usages

05/22/2020
There are now two dash python files; stock_dash.py is the stock price, candlestick and volume dashboard. A new file, stock_forecast.py contains forecast of up to 100 days from latest trading date.

To run forecast dash app, run in terminal the following:

python stock_forecast.py

Then in the web browser address bar enter localhost:8050.

Select a company from the dropdown and hit submit. After a few seconds, the graph updates, containing the historic stock price data, the forecast values, upper bound interval and lower bound interval.

04/14/2020
Explored Dash (Plotly) library to create simple dashboard to display candlestick graph of stock price up to a year. Includes open and close prices. Red indicates drop in price on the day, green indicates an increase in price.

To run the stock dashboard dash app, run in terminal the following:

python stock_dash.py

Then in the web browser address bar enter localhost:8050 (where 8050 is the port number to run stock_dash.py.)

About

Project to predict stock prices with Recurrent Neural Network in TensorFlow with client input as web application (with Flask).

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages