Rely

Welcome. Here lies Rely - a web app I created that will enable you to forecast today's stock market.

Introduction

Rely is a Flask web application that implements four machine learning algorithm to predict stock prices. Namely, those stocks are Twitter, Verizon, Facebook, Tesla, and Google. You can forecast up to five days in the future using the below machine learning algorithms:

• Linear Regression Algorithm
• K Fold Cross-Validation Algorithm
• Support Vector Regression Algorithm
• Naive Bayes Classification Algorithm

This application has been deployed so it can be accessed here: http://45.56.102.63/

Usage Instructions

Rely is an open source project. If you are interested in using the software for your own development, clone it from above with HTTPS or paste below into your terminal:

git clone https://github.com/baldeosinghm/rely.git

This tool uses Quandl's Time-Series API to retrieve historical stock prices. Currently, my API token is being used. However, you should replace it with one acquired via Quandl if you are to make any changes or avoid API call restrictions. This can be done by navigating to the generate_csv.py module in the src folder and replacing the portion below entitled API_token with yours.

endpoint = requests.get("https://www.quandl.com/api/v3/datasets/WIKI/" + stock + "/data.json?api_key=API_token")

If you would like to manipulate a different dataset refer to Quandl's API documentation here.

Install Dependencies

Rely is only operational when certain dependencies are installed. It is highly recommended that you create a virtual environment before any changes to the code are made. Once inside the repository, install pipenv, a dependency manager for Python projects.

Install pipenv in the terminal:

pip install --user pipenv

Navigate to rely/flaskapp/ and start virtual environment:

pipenv shell

Install app dependencies:

pipenv install --dev

Run Application

Navigate to the inside of /flaskapp and run:

python relyApp.py

Paste localhost:5000 into browser to run application.

Evaluation

To evaluate the quality and performance of the algorithms you can refer to the evaluation folder. Simply paste any of the given functions at the bottom of it's respective algorithm's module. The output will be printed in the terminal while the local server runs. An example is provided in the evaluation folder.