Amazon Machine Learning Project

⚠️ This project is archived and no longer actively maintained.

A Java-based project demonstrating Amazon Machine Learning (Amazon ML) capabilities, including data generation for machine learning experiments and real-time prediction using AWS ML services.

📋 Table of Contents

• Overview
• Features
• Prerequisites
• Installation
• Usage
• Project Structure
• Components
• Data Generation
• Machine Learning Predictions
• Configuration
• Contributing
• License

🎯 Overview

This project provides tools and examples for working with Amazon Machine Learning services. It includes:

• Data Generator: Creates synthetic web log data for ML training and testing
• ML Sample: Demonstrates real-time prediction using AWS ML endpoints
• Spark Examples: Code samples for Apache Spark integration

Originally created to support the blog post: Test Drive: AWS Machine Learning + Redshift

✨ Features

• 🔄 Synthetic Data Generation: Generate realistic web log data with demographic variations
• 🤖 Real-time ML Predictions: Connect to AWS ML endpoints for live predictions
• 📊 Data Processing: Handle CSV data with customizable delimiters
• 🌍 Geographic Variations: Include state-based pricing multipliers
• 📅 Temporal Patterns: Seasonal and time-based data variations
• 🔧 Maven Integration: Easy dependency management and build process

🛠 Prerequisites

• Java 8 or higher
• Maven 3.3 or higher
• AWS Account with Machine Learning service access
• AWS Credentials configured (via AWS CLI or environment variables)

📦 Installation

1. Clone the repository

   git clone https://github.com/julien/amazon-machine-learning.git
   cd amazon-machine-learning

2. Build the project

   mvn clean compile

3. Configure AWS credentials

   aws configure
   # Or set environment variables:
   export AWS_ACCESS_KEY_ID=your_access_key
   export AWS_SECRET_ACCESS_KEY=your_secret_key
   export AWS_DEFAULT_REGION=eu-west-1

🚀 Usage

Data Generation

Generate synthetic web log data:

# Compile and run the generator
javac -cp ".:lib/*" Generator.java
java -cp ".:lib/*" org.julien.datastuff.Generator

This will create data-batch-prediction.txt with 1000 synthetic records.

Machine Learning Predictions

Make predictions using an existing ML model:

# Compile the ML sample
javac -cp ".:lib/*" src/org/julien/datastuff/MLSample.java

# Run predictions (replace MODEL_ID with your actual model ID)
java -cp ".:lib/*:src" org.julien.datastuff.MLSample MODEL_ID

�� Project Structure

amazon-machine-learning/
├── README.md                 # This file
├── pom.xml                   # Maven configuration
├── Generator.java            # Data generation utility
├── data-batch-prediction.txt # Generated sample data
├── dist.all.last            # Last names dataset
├── dist.female.first        # Female first names dataset
├── dist.male.first          # Male first names dataset
├── US_States.txt            # US states dataset
└── src/
    └── org/julien/datastuff/
        └── MLSample.java     # ML prediction example

🔧 Components

Data Generator (Generator.java)

Generates synthetic web log data with the following features:

• Demographic Data: Names, gender, age, location
• Temporal Data: Day of year, hour, minutes
• Purchase Data: Number of items, basket price
• Geographic Multipliers: State-based pricing variations
• Seasonal Patterns: Holiday season multipliers

Generated Data Format:

LastName,FirstName,Gender,State,Age,DayOfYear,Hour,Minute,Items,PurchaseAmount

ML Sample (MLSample.java)

Demonstrates real-time prediction using AWS Machine Learning:

• Model Discovery: Lists available ML models
• Endpoint Connection: Connects to real-time prediction endpoints
• Prediction Requests: Sends data for prediction
• Response Processing: Handles prediction results

📊 Data Generation

The data generator creates realistic e-commerce data with the following characteristics:

Demographic Variations

• Gender-based pricing: Female customers have 25% higher average purchase values
• Age-based patterns: Customers aged 25-45 have 20% higher purchase values
• Geographic variations: Different states have different pricing multipliers

Geographic Multipliers

• California, Florida: 1.5x multiplier
• New York: 1.75x multiplier
• District of Columbia: 2.0x multiplier

Seasonal Patterns

• Holiday season (Nov-Dec): 1.5x to 2.5x multipliers
• Peak shopping periods: Enhanced purchase values

🤖 Machine Learning Predictions

The ML sample demonstrates:

1. Model Discovery: Find and list available ML models
2. Endpoint Connection: Connect to real-time prediction endpoints
3. Data Preparation: Format input data for prediction
4. Prediction Execution: Send requests and receive results
5. Performance Monitoring: Track request response times

Example Prediction Request

request.addRecordEntry("age", "32")
       .addRecordEntry("job", "management")
       .addRecordEntry("marital", "married")
       // ... additional features

⚙️ Configuration

AWS Configuration

• Region: Defaults to eu-west-1
• Credentials: Use AWS CLI or environment variables
• Permissions: Requires ML model access and prediction permissions

Data Files

The generator uses the following data files:

• dist.all.last: Last names database
• dist.male.first: Male first names database
• dist.female.first: Female first names database
• US_States.txt: US states list

🤝 Contributing

Since this project is archived, contributions are not being accepted. However, you can:

1. Fork the repository for your own use
2. Create issues for documentation improvements
3. Use the code as a reference for your own ML projects

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🔗 Related Resources

• AWS Machine Learning Documentation
• Original Blog Post
• Amazon ML API Reference

---

Note: This project was created for educational and demonstration purposes. The AWS Machine Learning service has been deprecated in favor of Amazon SageMaker. Consider migrating to SageMaker for production ML workloads.