import logging
import os
import sklearn
import numpy as np
from sklearn.neighbors import KNeighborsClassifier
from sklearn.neighbors import KNeighborsRegressor
from sklearn.model_selection import train_test_split
from supervised.algorithms.algorithm import BaseAlgorithm
from supervised.algorithms.sklearn import SklearnAlgorithm
from supervised.algorithms.registry import AlgorithmsRegistry
from supervised.algorithms.registry import (
BINARY_CLASSIFICATION,
MULTICLASS_CLASSIFICATION,
REGRESSION,
)
from supervised.utils.config import LOG_LEVEL
logger = logging.getLogger(__name__)
logger.setLevel(LOG_LEVEL)
from dtreeviz.trees import dtreeviz
class KNeighborsAlgorithm(SklearnAlgorithm):
algorithm_name = "k-Nearest Neighbors"
algorithm_short_name = "Nearest Neighbors"
def __init__(self, params):
super(KNeighborsAlgorithm, self).__init__(params)
logger.debug("KNeighborsAlgorithm.__init__")
self.library_version = sklearn.__version__
self.max_iters = 1
self.model = KNeighborsClassifier(
n_neighbors=params.get("n_neighbors", 3),
weights=params.get("weights", "uniform"),
algorithm="kd_tree",
n_jobs=-1,
)
def file_extension(self):
return "k_neighbors"
def fit(self, X, y, X_validation=None, y_validation=None, log_to_file=None):
if X.shape[0] > 1000:
X1, _, y1, _ = train_test_split(
X, y, train_size=1000, stratify=y, random_state=1234
)
self.model.fit(X1, y1)
else:
self.model.fit(X, y)
class KNeighborsRegressorAlgorithm(SklearnAlgorithm):
algorithm_name = "k-Nearest Neighbors"
algorithm_short_name = "Nearest Neighbors"
def __init__(self, params):
super(KNeighborsRegressorAlgorithm, self).__init__(params)
logger.debug("KNeighborsRegressorAlgorithm.__init__")
self.library_version = sklearn.__version__
self.max_iters = 1
self.model = KNeighborsRegressor(
n_neighbors=params.get("n_neighbors", 3),
weights=params.get("weights", "uniform"),
algorithm="ball_tree",
n_jobs=-1,
)
def file_extension(self):
return "k_neighbors"
def fit(self, X, y, X_validation=None, y_validation=None, log_to_file=None):
if X.shape[0] > 1000:
X1, _, y1, _ = train_test_split(X, y, train_size=1000, random_state=1234)
self.model.fit(X1, y1)
else:
self.model.fit(X, y)
knn_params = {"n_neighbors": [3, 5, 7], "weights": ["uniform", "distance"]}
default_params = {"n_neighbors": 5, "weights": "uniform"}
additional = {"max_rows_limit": 100000, "max_cols_limit": 100}
required_preprocessing = [
"missing_values_inputation",
"convert_categorical",
"scale",
"target_as_integer",
]
AlgorithmsRegistry.add(
BINARY_CLASSIFICATION,
KNeighborsAlgorithm,
knn_params,
required_preprocessing,
additional,
default_params,
)
AlgorithmsRegistry.add(
MULTICLASS_CLASSIFICATION,
KNeighborsAlgorithm,
knn_params,
required_preprocessing,
additional,
default_params,
)
AlgorithmsRegistry.add(
REGRESSION,
KNeighborsRegressorAlgorithm,
knn_params,
required_preprocessing,
additional,
default_params,
)
