게으른 우루루

import matplotlib.pyplot as plt

import mglearn

mglearn.plots.plot_knn_classification(n_neighbors=1)

plt.xlabel('feature 0')

plt.ylabel('feature 1')

plt.title('n_neighbors=1')

plt.legend(feature_list, loc=(1.01, 0.4))

plt.show()

n_neighbors=1 일 때 algorithm

mglearn.plots.plot_knn_classification(n_neighbors=3)

plt.xlabel('feature 0')

plt.ylabel('feature 1')

plt.title('n_neighbors=3')

plt.legend(feature_list, loc=(1.01, 0.4))

plt.show()

n_neighbors=1 일 때 algorithm

from mglearn.datasets import make_forge

x, y = make_forge()

_, axes = plt.subplots(1, 3)

for i, ax in zip([1, 3, 9], axes.ravel()):

    knn = KNeighborsClassifier(i)

    knn.fit(x, y)

    mglearn.plots.plot_2d_separator(knn, x, fill=True,

                                    eps=0.5, alpha=0.5, ax=ax)

    mglearn.discrete_scatter(x[:, 0], x[:, 1],

                             y=y, ax=ax)

    ax.set_title('k={}'.format(i))

    ax.set_xlabel('feature 0')

    ax.set_ylabel('feature 1')

axes[0].legend(loc=(0.01, 1.01))

plt.show()

n_neighbors 값에 따른 k-NN모델이 만든 decision boundary

from sklearn.datasets import load_wine

import pandas as pd

wine = load_wine()

print(wine.keys())

# dict_keys(['data', 'target', 'target_names', 'DESCR', 'feature_names'])

for i in range(len(wine.feature_names)):

    print('{}:{}'.format(i, wine.feature_names[i]))

# 0:alcohol

# 1:malic_acid

# 2:ash

# 3:alcalinity_of_ash

# 4:magnesium

# 5:total_phenols

# 6:flavanoids

# 7:nonflavanoid_phenols

# 8:proanthocyanins

# 9:color_intensity

# 10:hue

# 11:od280/od315_of_diluted_wines

# 12:proline    

wine_df = pd.DataFrame(wine.data, columns=range(len(wine.feature_names)))

pd.scatter_matrix(wine_df, # dataframe

                  c=wine.target, # color

                  hist_kwds={'bins':30}, # hist kwords

                  s=10, # size

                  alpha=0.5, # alpha

                  marker='o') # marker

plt.show()

wine의 scatter_plot

from sklearn.neighbors import KNeighborsClassifier

from sklearn.model_selection import train_test_split

x_train, x_test, y_train, y_test =\

  train_test_split(wine.data, wine.target, stratify=wine.target, # stratify = 계층화

                   random_state=0, test_size=0.3)  

knn = KNeighborsClassifier(n_neighbors=3, n_jobs=-1) # jobs =  사용할 코어의 수, -1 => 모든 코어

knn.fit(x_train, y_train)

print('{:.3f}'.format(knn.score(x_test, y_test)))

# 0.648

from sklearn.datasets import load_wine

import matplotlib.pyplot as plt

wine = load_wine()

x_train, x_test, y_train, y_test = \

  train_test_split(wine.data, wine.target, stratify=wine.target,

                   test_size=0.3, random_state=0)

train_list = []

test_list = []

n_range = range(1, 20)

for i in n_range:

    knn = KNeighborsClassifier(n_neighbors=i, n_jobs=-1)

    knn.fit(x_train, y_train)

    tr_score = knn.score(x_train, y_train)

    te_score = knn.score(x_test, y_test)

    train_list.append(tr_score)

    test_list.append(te_score)

plt.plot(n_range, train_list, color='red', ls='--', lw=2, label='train accuracy') # ls = linestyle, lw= linewidth

plt.plot(n_range, test_list, color='green', lw=2, label='test accuracy')

plt.xticks(n_range)

plt.xlabel('n_neighbors')

plt.ylabel('accuracy')

plt.legend() 

plt.show()

n_neighbors 변화에 따른 훈련 정확도와 테스트 정확도