This repository provides a clean, reproducible benchmarking pipeline comparing classical Support Vector Machines (SVM) with Quantum Support Vector Machines (QSVM) using Qiskit Machine Learning. The project is designed for students, educators, and researchers who want to understand when quantum kernel methods help, when they fail, and how preprocessing such as PCA affects feasibility.

This project demonstrates three regimes of quantum machine learning performance:

1. Quantum Advantage
   QSVM outperforms classical SVM on structured parity datasets.

2. Quantum Neutrality
   QSVM performs similarly to classical SVM after PCA reduces dimensionality.

3. Quantum Disadvantage
   QSVM becomes infeasible or unstable on high-dimensional real-world data without preprocessing.

All experiments run entirely on simulators. No quantum hardware access is required.

``` src/ phase1/ Classical SVM experiments (Parity, Breast Cancer, Breast Cancer PCA) phase2/ QSVM experiments on Parity datasets phase3/ QSVM experiments on Breast Cancer datasets utils/ Shared loaders, evaluators, loggers, and visualization tools

config/ YAML configuration for dataset selection data/ Public datasets (parity and breast cancer) plots/ Auto-generated plots (ignored by git) results.csv Logged experiment results ```

```cmd git clone https://github.com/your-username/classical-vs-quantum-svm.git cd classical-vs-quantum-svm ```

```cmd python -m venv .venv .venv\Scripts\activate ```

```cmd pip install -r requirements.txt ```

```cmd python -m src.phase1.SVM_Parity python -m src.phase2.QSVM_Parity python -m src.phase1.SVM_BreastCancer python -m src.phase1.SVM_BreastCancer_PCA python -m src.phase3.QSVM_BreastCancer python -m src.phase3.QSVM_BreastCancer_PCA ```

• Classical SVM struggles.
• QSVM performs significantly better.
• This is a clear example of quantum advantage.

• Classical SVM performs well.
• QSVM is not feasible due to dimensionality.
• This demonstrates quantum disadvantage.

• Both classical SVM and QSVM run successfully.
• Accuracy is comparable.
• This demonstrates quantum neutrality.

• Understand when quantum kernels help and when they fail.
• See how PCA enables QSVM feasibility on real datasets.
• Compare classical and quantum models on equal footing.
• Learn reproducible ML and QML experiment design.
• Run all experiments on Windows using Qiskit simulators.

This project is intended for educational use and is fully simulator-based.