近傍成分分析の有無に伴う最近傍の比較

近傍成分分析の有無による最近傍(nearest neighbor)分類の比較の例です。

本来の特徴のユークリッド距離と、近傍成分分析によって学習した変換後のユークリッド距離を用いた時の最近傍分類によって与えれるクラスの決定範囲を図示します。後者の目的は、そのトレーニングセットの上で(確率的な)最近傍分類の精度を最大化する線形変換を見つけることです。

# License: BSD 3 clause

import matplotlib.pyplot as plt
from matplotlib.colors import ListedColormap
from sklearn import datasets
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.neighbors import KNeighborsClassifier, NeighborhoodComponentsAnalysis
from sklearn.pipeline import Pipeline
from sklearn.inspection import DecisionBoundaryDisplay


n_neighbors = 1

dataset = datasets.load_iris()
X, y = dataset.data, dataset.target

# we only take two features. We could avoid this ugly
# slicing by using a two-dim dataset
X = X[:, [0, 2]]

X_train, X_test, y_train, y_test = train_test_split(
    X, y, stratify=y, test_size=0.7, random_state=42
)

h = 0.05  # step size in the mesh

# Create color maps
cmap_light = ListedColormap(["#FFAAAA", "#AAFFAA", "#AAAAFF"])
cmap_bold = ListedColormap(["#FF0000", "#00FF00", "#0000FF"])

names = ["KNN", "NCA, KNN"]

classifiers = [
    Pipeline(
        [
            ("scaler", StandardScaler()),
            ("knn", KNeighborsClassifier(n_neighbors=n_neighbors)),
        ]
    ),
    Pipeline(
        [
            ("scaler", StandardScaler()),
            ("nca", NeighborhoodComponentsAnalysis()),
            ("knn", KNeighborsClassifier(n_neighbors=n_neighbors)),
        ]
    ),
]

for name, clf in zip(names, classifiers):

    clf.fit(X_train, y_train)
    score = clf.score(X_test, y_test)

    _, ax = plt.subplots()
    DecisionBoundaryDisplay.from_estimator(
        clf,
        X,
        cmap=cmap_light,
        alpha=0.8,
        ax=ax,
        response_method="predict",
        plot_method="pcolormesh",
        shading="auto",
    )

    # Plot also the training and testing points
    plt.scatter(X[:, 0], X[:, 1], c=y, cmap=cmap_bold, edgecolor="k", s=20)
    plt.title("{} (k = {})".format(name, n_neighbors))
    plt.text(
        0.9,
        0.1,
        "{:.2f}".format(score),
        size=15,
        ha="center",
        va="center",
        transform=plt.gca().transAxes,
    )

plt.show()