Bhomic-OGJ/QEFS ? reverse-engineered prompt

Build me a notebook based project for the QEFS feature selection workflow, where I can load the processed UCI datasets, run the quantum evolutionary feature selector, and compare the selected features with a normal classifier based fitness score. I want it to work both on a local simulator and, if I add my IBM Quantum key, on real IBM hardware too.

Please make the notebooks easy to run from start to finish, with clear cells for data loading, preprocessing, running the algorithm, and showing results like selected features, fitness over generations, and any drift or robustness plots. Use the same general settings across datasets so the experiments are repeatable, and include sensible defaults for the hyperparameters mentioned in the repo. If anything needs newer Qiskit guidance or current IBM setup steps, look up the latest docs online.