config.py

import os
import sys

from sklearn.ensemble import RandomForestClassifier
from sklearn.discriminant_analysis import LinearDiscriminantAnalysis

# Path to directory with sslearn folder if not in the PATH already.
sslearn_root = os.path.dirname(__file__)
sys.path.append(sslearn_root)

# Import algorithm to test.
from sslearn.models.sla._customs import BinarySLARandomForestClassifier  # noqa
from sslearn.models.lda import SemiSupervisedLinearDiscriminantAnalysis  # noqa
from sslearn.models.lgc import LGC  # noqa
from sslearn.models.rf import (
    SemiSupervisedRandomForest,
    RandomForest,
)  # noqa
from sslearn.models.clustering._customs import SSClusteringClassifier  # noqa


configs = {
    "sla": {
        "model_cls": BinarySLARandomForestClassifier,
        "baseline_cls": RandomForestClassifier,
        "model_inits": {
            "n_estimators": 100,
            "adaptive": True,
            "margin_mode": "soft",
            "max_iter": 150,
        },
        "baseline_inits": {"n_estimators": 100,},
    },
    "lda": {
        "model_cls": SemiSupervisedLinearDiscriminantAnalysis,
        "baseline_cls": LinearDiscriminantAnalysis,
    },
    "lgc": {"model_cls": LGC},
    "rf": {"model_cls": SemiSupervisedRandomForest, "baseline_cls": RandomForest},
    "clustering": {
        "model_cls": SSClusteringClassifier,
        "baseline_cls": RandomForestClassifier,
        "model_inits": {
            "eps": 0.1,
            "min_points": 10,
            "random_state": 42,
            "use_border_points": True,
            "n_estimators": 100,
        },
        "baseline_inits": {"random_state": 42, "n_estimators": 100,},
    },
}