1. 程式人生 > >15 SVM - 程式碼案例二 - 鳶尾花資料不同分類器效果比較

15 SVM - 程式碼案例二 - 鳶尾花資料不同分類器效果比較

SVM的章節已經講完,具體內容請參考:《01 SVM - 大綱

14 SVM - 程式碼案例一 - 鳶尾花資料SVM分類

回顧案例一中的標頭檔案:

import numpy as np
import pandas as pd
import matplotlib as mpl
import matplotlib.pyplot as plt
import warnings
from sklearn import svm#svm匯入
from sklearn.svm import SVC
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score
from sklearn.exceptions import ChangedBehaviorWarning
案例二 - 鳶尾花資料不同分類器效果比較
常規操作:

1、標頭檔案引入SVM相關的包
2、防止中文亂碼
3、去警告
4、讀取資料
5、資料分割訓練集和測試集 6:4

import numpy as np
import pandas as pd
import matplotlib as mpl
import matplotlib.pyplot as plt
import warnings
from sklearn.svm import SVC
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score
from sklearn.linear_model import LogisticRegression,RidgeClassifier
from sklearn.neighbors import KNeighborsClassifier


## 設定屬性防止中文亂碼
mpl.rcParams['font.sans-serif'] = [u'SimHei']
mpl.rcParams['axes.unicode_minus'] = False

warnings.filterwarnings('ignore', category=ChangedBehaviorWarning)

## 讀取資料
# 'sepal length', 'sepal width', 'petal length', 'petal width'
iris_feature = u'花萼長度', u'花萼寬度', u'花瓣長度', u'花瓣寬度'
path = './datas/iris.data'  # 資料檔案路徑
data = pd.read_csv(path, header=None)
x, y = data[list(range(4))], data[4]
y = pd.Categorical(y).codes #把文字資料進行編碼,比如a b c編碼為 0 1 2
x = x[[0, 1]]

## 資料分割
x_train, x_test, y_train, y_test = train_test_split(x, y, random_state=28, train_size=0.6)

資料SVM分類器構建:
svm = SVC(C=1, kernel='linear')

## 模型訓練
svm.fit(x_train, y_train)

svm.intercept_


Linear分類器構建:

RidgeClassifier(): ridge是為了解決特徵大於樣本,而導致分類效果較差的情況,而提出的。
svm有一個重要的瓶頸——當特徵數大於樣本數的時候,效果變差。

lr = LogisticRegression()
rc = RidgeClassifier()
knn = KNeighborsClassifier()

## 模型訓練
lr.fit(x_train, y_train)
rc.fit(x_train, y_train)
knn.fit(x_train, y_train)


效果評估:

svm_score1 = accuracy_score(y_train, svm.predict(x_train))
svm_score2 = accuracy_score(y_test, svm.predict(x_test))

lr_score1 = accuracy_score(y_train, lr.predict(x_train))
lr_score2 = accuracy_score(y_test, lr.predict(x_test))

rc_score1 = accuracy_score(y_train, rc.predict(x_train))
rc_score2 = accuracy_score(y_test, rc.predict(x_test))

knn_score1 = accuracy_score(y_train, knn.predict(x_train))
knn_score2 = accuracy_score(y_test, knn.predict(x_test))
畫圖 - 鳶尾花資料不同分類器準確率比較:
x_tmp = [0,1,2,3]
y_score1 = [svm_score1, lr_score1, rc_score1, knn_score1]
y_score2 = [svm_score2, lr_score2, rc_score2, knn_score2]

plt.figure(facecolor='w')
plt.plot(x_tmp, y_score1, 'r-', lw=2, label=u'訓練集準確率')
plt.plot(x_tmp, y_score2, 'g-', lw=2, label=u'測試集準確率')
plt.xlim(0, 3)
plt.ylim(np.min((np.min(y_score1), np.min(y_score2)))*0.9, np.max((np.max(y_score1), np.max(y_score2)))*1.1)
plt.legend(loc = 'lower right')
plt.title(u'鳶尾花資料不同分類器準確率比較', fontsize=16)
plt.xticks(x_tmp, [u'SVM', u'Logistic', u'Ridge', u'KNN'], rotation=0)
plt.grid(b=True)
plt.show()


畫圖比較分類結果:

N = 500
x1_min, x2_min = x.min()
x1_max, x2_max = x.max()

t1 = np.linspace(x1_min, x1_max, N)
t2 = np.linspace(x2_min, x2_max, N)
x1, x2 = np.meshgrid(t1, t2)  # 生成網格取樣點
grid_show = np.dstack((x1.flat, x2.flat))[0] # 測試點

獲取各個不同演算法的測試值:

svm_grid_hat = svm.predict(grid_show)
svm_grid_hat = svm_grid_hat.reshape(x1.shape)  # 使之與輸入的形狀相同

lr_grid_hat = lr.predict(grid_show)
lr_grid_hat = lr_grid_hat.reshape(x1.shape)  # 使之與輸入的形狀相同

rc_grid_hat = rc.predict(grid_show)
rc_grid_hat = rc_grid_hat.reshape(x1.shape)  # 使之與輸入的形狀相同

knn_grid_hat = knn.predict(grid_show)
knn_grid_hat = knn_grid_hat.reshape(x1.shape)  # 使之與輸入的形狀相同

畫圖:

cm_light = mpl.colors.ListedColormap(['#A0FFA0', '#FFA0A0', '#A0A0FF'])
cm_dark = mpl.colors.ListedColormap(['g', 'r', 'b'])
plt.figure(facecolor='w', figsize=(14,7))

1、鳶尾花SVM特徵分類:

plt.subplot(221)
## 區域圖
plt.pcolormesh(x1, x2, svm_grid_hat, cmap=cm_light)
## 所以樣本點
plt.scatter(x[0], x[1], c=y, edgecolors='k', s=50, cmap=cm_dark)      # 樣本
## 測試資料集
plt.scatter(x_test[0], x_test[1], s=120, facecolors='none', zorder=10)     # 圈中測試集樣本
## lable列表
plt.xlabel(iris_feature[0], fontsize=13)
plt.ylabel(iris_feature[1], fontsize=13)
plt.xlim(x1_min, x1_max)
plt.ylim(x2_min, x2_max)
plt.title(u'鳶尾花SVM特徵分類', fontsize=16)
plt.grid(b=True, ls=':')
plt.tight_layout(pad=1.5)

2、鳶尾花Logistic特徵分類:

plt.subplot(222)
## 區域圖
plt.pcolormesh(x1, x2, lr_grid_hat, cmap=cm_light)
## 所以樣本點
plt.scatter(x[0], x[1], c=y, edgecolors='k', s=50, cmap=cm_dark)      # 樣本
## 測試資料集
plt.scatter(x_test[0], x_test[1], s=120, facecolors='none', zorder=10)     # 圈中測試集樣本
## lable列表
plt.xlabel(iris_feature[0], fontsize=13)
plt.ylabel(iris_feature[1], fontsize=13)
plt.xlim(x1_min, x1_max)
plt.ylim(x2_min, x2_max)
plt.title(u'鳶尾花Logistic特徵分類', fontsize=16)
plt.grid(b=True, ls=':')
plt.tight_layout(pad=1.5)

3、鳶尾花Ridge特徵分類:

plt.subplot(223)
## 區域圖
plt.pcolormesh(x1, x2, rc_grid_hat, cmap=cm_light)
## 所以樣本點
plt.scatter(x[0], x[1], c=y, edgecolors='k', s=50, cmap=cm_dark)      # 樣本
## 測試資料集
plt.scatter(x_test[0], x_test[1], s=120, facecolors='none', zorder=10)     # 圈中測試集樣本
## lable列表
plt.xlabel(iris_feature[0], fontsize=13)
plt.ylabel(iris_feature[1], fontsize=13)
plt.xlim(x1_min, x1_max)
plt.ylim(x2_min, x2_max)
plt.title(u'鳶尾花Ridge特徵分類', fontsize=16)
plt.grid(b=True, ls=':')
plt.tight_layout(pad=1.5)

4、鳶尾花KNN特徵分類:

plt.subplot(224)
## 區域圖
plt.pcolormesh(x1, x2, knn_grid_hat, cmap=cm_light)
## 所以樣本點
plt.scatter(x[0], x[1], c=y, edgecolors='k', s=50, cmap=cm_dark)      # 樣本
## 測試資料集
plt.scatter(x_test[0], x_test[1], s=120, facecolors='none', zorder=10)     # 圈中測試集樣本
## lable列表
plt.xlabel(iris_feature[0], fontsize=13)
plt.ylabel(iris_feature[1], fontsize=13)
plt.xlim(x1_min, x1_max)
plt.ylim(x2_min, x2_max)
plt.title(u'鳶尾花KNN特徵分類', fontsize=16)
plt.grid(b=True, ls=':')
plt.tight_layout(pad=1.5)
plt.show()

比較分類結果

16 SVM - 程式碼案例三 - 不同SVM核函式效果比較