Quick Start

Installation

pip install git+https://github.com/tungduy1704/Qmes.git

or, for a development install:

git clone https://github.com/tungduy1704/Qmes.git
cd Qmes
pip install -e .

Note

Rebuilding the meta-dataset from UCI sources (the offline training pipeline in Qmes/data/) requires the optional data extra: pip install -e ".[data]". Inference and the default recommenders do not need it.

Classification example

from sklearn.datasets import load_breast_cancer
from Qmes import get_extractor, load_default_recommender, recommend

X, y = load_breast_cancer(return_X_y=True)

extractor = get_extractor("classification")
recommender = load_default_recommender("classification")
result = recommend(X, y, extractor=extractor, recommender=recommender)

print("Top circuits:", result["top_k"])
# Top circuits: ['unit', 'RY', 'HERx']
print("Full ranking:", result["ranking"])
# Full ranking: ['unit', 'RY', 'HERx', 'SRx', 'RY_CX', 'HD', 'ZFM']
print("Vote counts:", result['votes'])
# Vote counts: {'unit': 6, 'SRx': 3, 'RY': 5, 'HERx': 4, 'RY_CX': 2, 'ZFM': 0, 'HD': 1}

Regression example

from sklearn.datasets import load_diabetes
from Qmes import get_extractor, load_default_recommender, recommend

X, y = load_diabetes(return_X_y=True)

extractor = get_extractor("regression")
recommender = load_default_recommender("regression")

result = recommend(X, y, extractor=extractor, recommender=recommender)

print("Top circuits:", result["top_k"])
# Top circuits: ['RY', 'unit', 'HERx']
print("Full ranking:", result["ranking"])
# Full ranking: ['RY', 'unit', 'HERx', 'HD', 'RY_CX', 'SRx', 'ZFM']
print("Vote counts:", result['votes'])
# Vote counts: {'unit': 5, 'SRx': 1, 'RY': 6, 'HERx': 4, 'RY_CX': 2, 'ZFM': 0, 'HD': 3}

Understanding the output

Key	Type	Description
ranking	list[str]	All 7 circuits sorted by votes in descending order
top_k	list[str]	First top_k elements of ranking (default: 3)
votes	dict[str, int]	Raw vote count per circuit
meta_features	np.ndarray	Complexity features extracted from your dataset

Available circuits

from Qmes import get_circuit_names
print(get_circuit_names())
# ['unit', 'SRx', 'RY', 'HERx', 'RY_CX', 'ZFM', 'HD']