Tiny Project: Linear Algebra & CPU Performance Regression

Overview

This project implements:

• A linear algebra library (Vector, Matrix, LinearSystem classes)
• CPU performance prediction using linear regression on UCI's Computer Hardware dataset

Requirements

• C++ compiler (g++/MinGW)
• Make utility

Features

Part A: Linear Algebra Library

• Vector and Matrix operations
• Linear system solver (Gaussian elimination)
• Positive definite system solver (Conjugate gradient)

Part B: Linear Regression

• Predicts CPU performance (PRP) using 6 hardware features
• 80/20 train-test split
• Reports RMSE metrics

Getting Started

Dataset Setup

1. Download the Computer Hardware dataset from UCI: Computer Hardware Dataset
2. Place the machine.data file in the data/ folder

Building the Project

Using Makefile

# On Linux/Mac:
make

# On Windows:
mingw32-make

# Run the program:
./tinyProject

Manual Compilation

g++ -std=c++11 -Iinclude src/*.cpp main.cpp -o tinyProject
./tinyProject

Example Output

Regression coefficients:
MYCT: -0.012, MMIN: 0.005, MMAX: 0.003
CACH: 0.821, CHMIN: -0.104, CHMAX: 0.542

Training RMSE: 25.34
Testing RMSE: 28.71

Project Structure

tinyProject/
├── include/       # Header files
├── src/          # Source code
├── data/         # Dataset
└── Makefile      # Build script

Troubleshooting

• On Windows: Use mingw32-make instead of make
• Ensure machine.data exists in data/ folder
• Requires C++11 support

Links

• GitHub: https://github.com/1N2H

---