ML Forecast Application

This is python web application build using Streamlit for predicting stock prices using Facebook's forecasting algorithm Prophet

Please refer to the docs folder in which there is a Python based Jupyter note book detailing all relevant comments pertaining to the the individual functions this application is developed upon

This has been exported to a PDF file for ease of reading at Stock Prediction Application

This application can be executed via two methods:

1. Using native local host execution via Streamlit's inbuilt web server protocol
2. Using docker

Application Runtime

Locally

Prerequisites as per the text file

Install Python Libraries

pip3 install -r requirements.txt

Run application locally

steamlit run src/main.py

Expected output:

Docker build and execution

Prerequisites:

• Docker
• docker-compose

Build Docker image in current directory

The docker logic has been developed as follows:

• Multi-stage build. In the first stage all necessary Python packages are installed inside a virtual environment
• Create a non-root user with the correct user permissions. Then copy the Python libraries from the build image into the runtime image.
• Mount local folders from your computer to the Docker container. Develop the app while using Docker and save changes.

A docker compose file has been utilised for convenience of application loading

To execute the build of the docker image build via the below command:

docker-compose build

Expected Output:

Run docker image with port 8501 and volumes

docker run -it --rm --port '8501:8501' -v '$(pwd)/project:/usr/scr/app/project:delegated' latest

Run Application

Start the Docker container in the background with docker-compose

docker-compose up -d

System Requirements

From local testing, to successfully run this application the minimum spec is:

• 3 vCPU
• 10 GB RAM

*** Note: This is primarily due to system requirements for running PyStan ***