GPU-Crypto-Policy-Analysis 📉🏛️

Analyzing the Hidden Drivers of GPU Prices: Cryptocurrency, Government Regulation, and Military Defense Spending.

📋 Project Overview

This project explores the complex relationship between GPU hardware prices and three distinct market forces:

1. Cryptocurrency Markets: Bitcoin & Ethereum mining demand.
2. Government Regulation: The impact of crypto bans, SEC approvals, and global policy.
3. Military & Defense: The "hidden driver" of AI/Cybersecurity defense procurement.

Using a Knowledge Discovery in Databases (KDD) approach, we integrated real-world data from 2019-2025 to prove that while Crypto drives volatility, Government and Military factors act as critical "regime shifters" in the market.

---

🚀 Key Features

• Real-World Data Pipeline: Integrates Yahoo Finance (Stocks/Crypto), Federal Register (Regulations), and USASpending.gov (Defense Budgets).
• 28 Engineered Features: Includes Technical Indicators (RSI, MACD), Interaction Terms (Gov_Crypto_Interaction), and Volatility metrics.
• Regime Detection: Identifies distinct market states (e.g., "Regulated Bull Market", "Defense-Driven Demand").
• Automated Preprocessing: Full Python pipeline from raw API data to ML-ready scaled datasets.

---

📂 Repository Structure

├── code/
│   ├── 01_collect_real_crypto_gpu_data.py   # Fetches Crypto & Stock Data
│   ├── 02_collect_real_gov_military_data.py # Adds Gov/Military Events
│   ├── 03_integrate_real_data.py            # Merges & Cleans Data
│   └── 04_preprocessing_real_data.py        # Feature Engineering & Scaling
├── data/
│   ├── raw/                                 # Raw CSVs
│   └── processed/                           # Final ML-ready datasets
├── reports/
│   ├── Section_1_Introduction.md            # Problem Statement & Research Questions
│   └── Section_2_Approach.md                # Methodology & Data Dictionary
└── outputs/
    └── feature_engineering_documentation.json # Full feature list

---

📊 Data Sources

• Cryptocurrency: Bitcoin (BTC), Ethereum (ETH) via yfinance.
• Stocks: NVIDIA (NVDA), AMD (AMD), Intel (INTC).
• Government: Major regulatory events (EU MiCA, US GENIUS Act).
• Military: Defense AI Budget Authority & Geopolitical Conflict Events.

---

This repository represents the Data Architect phase of the project:

• ✅ Data Collection: Complete
• ✅ Preprocessing: Complete
• ✅ Feature Engineering: Complete
• ✅ Data Mining & Analysis: Complete

---

🔬 Data Analysis & Model Implementation

Correlation Analysis

• Objective: Identify relationships between GPU prices and market factors
• Method: Pearson correlation matrix with 18 key features
• Key Findings:
  • Extremely strong correlation between NVDA and GPU Index (r = 0.9960)
  • Very strong BTC-ETH correlation (r = 0.8118)
  • Moderate GPU-Crypto correlation (r = 0.9135)
  • Intel shows negative correlation with GPU markets
• Outputs:
  • correlation_heatmap.png: Visual correlation matrix
  • correlation_analysis_results.txt: Detailed correlation results

Regression Analysis (Ridge Regression)

• Objective: Predict GPU_Stock_Index using normalized features
• Method: Ridge Regression with α=1.0, 80/20 train-test split
• Metrics:
  • R² Score: Explained variance of GPU prices
  • RMSE & MAE: Prediction error metrics
  • Feature coefficients: Impact of each feature
• Key Insights:
  • Negative military coefficient reveals complex market dynamics
  • Regulatory features significantly impact GPU prices
  • Crypto market features are strong predictors
• Outputs:
  • ridge_coefficients.png: Top 15 feature coefficients
  • regression_actual_vs_predicted.png: Model performance visualization
  • regression_analysis_results.txt: Complete regression metrics

Classification Analysis (Random Forest)

• Objective: Classify GPU_Price_Category (Low/Medium/High)
• Method: Random Forest with 100 trees, max_depth=10, stratified split
• Metrics:
  • Accuracy: Percentage of correct classifications
  • Precision & Recall: Class-specific performance
  • Feature Importance: Most predictive features
• Key Insights:
  • Government/military features rank high in importance
  • Model validates three-way market interaction hypothesis
  • High classification accuracy achieved
• Outputs:
  • confusion_matrix.png: Classification performance matrix
  • feature_importance_rf.png: Top 20 feature importances
  • classification_performance_by_class.png: Per-class metrics
  • classification_analysis_results.txt: Full classification report

Clustering Analysis (K-Means)

• Objective: Identify distinct market regimes
• Method: K-Means clustering with Elbow Method optimization
• Results:
  • Optimal clusters: k=2 (from silhouette and inertia analysis)
  • Cluster 0: High-Performance, High-Regulation Market
  • Cluster 1: Low-Performance, High-Regulation Market
• Key Insights:
  • Government policies create distinct market regimes
  • Military spending correlates with specific cluster characteristics
  • Time series shows regime evolution
• Outputs:
  • elbow_silhouette.png: Optimal k determination
  • clustering_pca.png: 2D cluster visualization
  • clusters_over_time.png: Temporal cluster distribution
  • cluster_radar_chart.png: Multi-dimensional cluster comparison
  • clustering_analysis_results.txt: Complete cluster profiles
  • data_with_clusters.csv: Dataset with cluster labels

---

This repository represents the complete project implementation:

• ✅ Data Collection: Complete
• ✅ Preprocessing: Complete
• ✅ Feature Engineering: Complete
• ✅ Correlation Analysis: Complete
• ✅ Regression Analysis: Complete
• ✅ Classification Analysis: Complete
• ✅ Clustering Analysis: Complete
• ✅ Ready for: Teammate 3