automl

[project]
name = "automl"
version = "0.1.0"
description = "AutoML experimentation environment"
requires-python = "==3.12.*"
dependencies = [
  "autogluon==1.5.0; platform_system != 'Windows'",
  "ray==2.52.1",
  "fastai==2.8.7",
  "statsforecast==2.0.1",
  "spacy==3.8.11",
  "flaml==2.5.0",
  "optuna==4.7.0",
  "ipykernel==7.2.0",
  "pytest==9.0.2",
  "ruff==0.15.2",
]

[tool.uv]
package = false

uv run python -c "import ray, fastai, flaml, optuna; print('ray', ray.__version__); print('fastai', fastai.__version__); print('flaml', flaml.__version__)"

uv run python -c "import autogluon; print('autogluon', autogluon.__version__)"

from autogluon.tabular import TabularPredictor

# 데이터 준비 (pandas DataFrame)
train_data = "train.csv"  # label 컬럼 포함

# 한 줄로 학습 (앙상블 스태킹 자동 수행)
predictor = TabularPredictor(label="target", eval_metric="accuracy").fit(
    train_data, time_limit=300  # 5분 제한
)

# 예측 및 리더보드
predictions = predictor.predict("test.csv")
leaderboard = predictor.leaderboard(silent=True)
print(leaderboard)

import optuna
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import cross_val_score
from sklearn.datasets import load_iris

X, y = load_iris(return_X_y=True)

def objective(trial):
    n_estimators = trial.suggest_int("n_estimators", 50, 300)
    max_depth = trial.suggest_int("max_depth", 3, 15)
    clf = RandomForestClassifier(n_estimators=n_estimators, max_depth=max_depth)
    return cross_val_score(clf, X, y, cv=5).mean()

study = optuna.create_study(direction="maximize")
study.optimize(objective, n_trials=50)
print(f"Best params: {study.best_params}, Best score: {study.best_value:.4f}")