Investor AI Website

Welcome to the stock price prediction project! This project leverages modern technologies like Airflow, Spark, MinIO, Java, React.js, Machine Learning and Docker to build a complete system, from data collection and processing to model building and deployment of a web interface that allows users to interact with and view prediction results.

Website Introduction

The project's website provides an intuitive interface for users to:

• View stock price information: Price charts, statistical indicators.
• Predict prices: Enter a stock symbol and a prediction period; the system will display the results.
• Evaluate the model: See model performance metrics (e.g., RMSE, MAE).

System Architecture

[System Architecture Image ( Not yet )]

The project is deployed based on a microservices architecture, with the following key components:

• Frontend: User interface (React.js).
• Backend (API): Provides APIs for the frontend and other services (Spring Boot - Java).
• Airflow: Manages the workflow of data collection, processing, and model training.
• Spark: Processing and machine learning model training* Handles large-scale data pr.
• MinIO: Stores data (data lake) and models.

How to Run the Project (Docker Compose)

The project is packaged using Docker and managed with Docker Compose, making deployment easy and consistent across all environments.

1. Install Docker and Docker Compose: Make sure you have Docker and Docker Compose installed on your machine. Installation instructions:

   • Docker: https://docs.docker.com/get-docker/
   • Docker Compose: https://docs.docker.com/compose/install/

2. Clone the Project:

   git clone https://github.com/ltdungg/Investor-AI-Website.git
   cd Investor-AI-Website

3. Environment for access:

   # POSTGRES
   POSTGRES_USER=airflow
   POSTGRES_PASSWORD=airflow
   POSTGRES_DB=airflow
   POSTGRES_STOCK_SCHEMA=stock
   
   # AIRFLOW
   AIRFLOW_USERNAME=airflow
   AIRFLOW_PASSWORD=airflow
   
   # MINIO
   MINIO_ROOT_USER=admin
   MINIO_ROOT_PASSWORD=admin123456789
   

4. Run Docker Compose:

   docker compose up --build -d

   This command will download the necessary images (if not already present), build containers, and start the services. The -d option runs the containers in detached mode (in the background).

5. Stop the Project

    docker compose down

Data Import (Volumes)

The project uses Docker Volumes to manage data, ensuring that data is not lost when containers are deleted.

1. Download volume from Google Drive for data:

   1. PostgreSQL volume: Click here to download
   2. MinIO volume: Not yet

2. Import Data into Volumes:

   1. Import PostgreSQL volume:
      • Select docker volume in Docker Desktop Volumes UI: investor-ai-website_postgres-db-volume
      • Select Import.
      • Select Local file and Browse your file .tar.gz you just downloaded from Google Drive.
      • Select Import.
   2. Import MinIO Volume:
      • Not yet.

Access Ports

Below are the default ports for the services in the project. You can change these ports in the docker-compose.yaml file.

Service	Port (Host)	Port (Container)	Description
Frontend	localhost	3000	User interface
Backend (API)	localhost	8000	Application API
Airflow	localhost	8080	Airflow Web UI
Spark	localhost	8081, 7077	Spark Master UI (8081),Spark Master (7077)
MinIO	localhost	9001, 9000	MinIO Console (9001), MinIO API (9000)
[Not yet]

For example: http://localhost:3000 to access the frontend, http://localhost:8080 to access Airflow.