鳶尾花(iris)數據集分析
Iris 鳶尾花數據集是一個經典數據集，在統計學習和機器學習領域都經常被用作示例。數據集內包含 3 類共 150 條記錄，每類各 50 個數據，每條記錄都有 4 項特征：花萼長度、花萼寬度、花瓣長度、花瓣寬度，可以通過這4個特征預測鳶尾花卉屬於（iris-setosa, iris-versicolour, iris-virginica）中的哪一品種。
據說在現實中，這三種花的基本判別依據其實是種子（因為花瓣非常容易枯萎）。
0 准備數據
下面對 iris 進行探索性分析，首先導入相關包和數據集：
# 導入相關包 import numpy as np import pandas as pd from pandas import plotting %matplotlib inline import matplotlib.pyplot as plt plt.style.use('seaborn') import seaborn as sns sns.set_style("whitegrid") from sklearn.linear_model import LogisticRegression from sklearn.model_selection import train_test_split from sklearn.preprocessing import LabelEncoder from sklearn.neighbors import KNeighborsClassifier from sklearn import svm from sklearn import metrics from sklearn.tree import DecisionTreeClassifier # 導入數據集 iris = pd.read_csv('F:\pydata\dataset\kaggle\iris.csv', usecols=[1, 2, 3, 4, 5]) 查看數據集信息：
iris.info()
<class 'pandas.core.frame.DataFrame'> RangeIndex: 150 entries, 0 to 149 Data columns (total 5 columns): SepalLengthCm 150 non-null float64 SepalWidthCm 150 non-null float64 PetalLengthCm 150 non-null float64 PetalWidthCm 150 non-null float64 Species 150 non-null object dtypes: float64(4), object(1) memory usage: 5.9+ KB 查看數據集的頭 5 條記錄：
iris.head()
1 探索性分析
先查看數據集各特征列的摘要統計信息：
iris.describe()
通過Violinplot 和 Pointplot，分別從數據分布和斜率，觀察各特征與品種之間的關系：
# 設置顏色主題 antV = ['#1890FF', '#2FC25B', '#FACC14', '#223273', '#8543E0', '#13C2C2', '#3436c7', '#F04864'] # 繪制 Violinplot f, axes = plt.subplots(2, 2, figsize=(8, 8), sharex=True) sns.despine(left=True) sns.violinplot(x='Species', y='SepalLengthCm', data=iris, palette=antV, ax=axes[0, 0]) sns.violinplot(x='Species', y='SepalWidthCm', data=iris, palette=antV, ax=axes[0, 1]) sns.violinplot(x='Species', y='PetalLengthCm', data=iris, palette=antV, ax=axes[1, 0]) sns.violinplot(x='Species', y='PetalWidthCm', data=iris, palette=antV, ax=axes[1, 1]) plt.show() 
# 繪制 pointplot f, axes = plt.subplots(2, 2, figsize=(8, 8), sharex=True) sns.despine(left=True) sns.pointplot(x='Species', y='SepalLengthCm', data=iris, color=antV[0], ax=axes[0, 0]) sns.pointplot(x='Species', y='SepalWidthCm', data=iris, color=antV[0], ax=axes[0, 1]) sns.pointplot(x='Species', y='PetalLengthCm', data=iris, color=antV[0], ax=axes[1, 0]) sns.pointplot(x='Species', y='PetalWidthCm', data=iris, color=antV[0], ax=axes[1, 1]) plt.show() 
生成各特征之間關系的矩陣圖：
g = sns.pairplot(data=iris, palette=antV, hue= 'Species') 
使用 Andrews Curves 將每個多變量觀測值轉換為曲線並表示傅立葉級數的系數，這對於檢測時間序列數據中的異常值很有用。
Andrews Curves 是一種通過將每個觀察映射到函數來可視化多維數據的方法。
plt.subplots(figsize = (10,8)) plotting.andrews_curves(iris, 'Species', colormap='cool') plt.show() 
下面分別基於花萼和花瓣做線性回歸的可視化：
g = sns.lmplot(data=iris, x='SepalWidthCm', y='SepalLengthCm', palette=antV, hue='Species') 
g = sns.lmplot(data=iris, x='PetalWidthCm', y='PetalLengthCm', palette=antV, hue='Species') 
最后，通過熱圖找出數據集中不同特征之間的相關性，高正值或負值表明特征具有高度相關性：
fig=plt.gcf()
fig.set_size_inches(12, 8) fig=sns.heatmap(iris.corr(), annot=True, cmap='GnBu', linewidths=1, linecolor='k', square=True, mask=False, vmin=-1, vmax=1, cbar_kws={"orientation": "vertical"}, cbar=True) 
從熱圖可看出，花萼的寬度和長度不相關，而花瓣的寬度和長度則高度相關。
2 機器學習
接下來，通過機器學習，以花萼和花瓣的尺寸為根據，預測其品種。
在進行機器學習之前，將數據集拆分為訓練和測試數據集。首先，使用標簽編碼將 3 種鳶尾花的品種名稱轉換為分類值（0, 1, 2）。
# 載入特征和標簽集 X = iris[['SepalLengthCm', 'SepalWidthCm', 'PetalLengthCm', 'PetalWidthCm']] y = iris['Species'] # 對標簽集進行編碼 encoder = LabelEncoder() y = encoder.fit_transform(y) print(y) [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2] 接着，將數據集以 7: 3 的比例，拆分為訓練數據和測試數據：
train_X, test_X, train_y, test_y = train_test_split(X, y, test_size = 0.3, random_state = 101) print(train_X.shape, train_y.shape, test_X.shape, test_y.shape) (105, 4) (105,) (45, 4) (45,)
檢查不同模型的准確性：
# Support Vector Machine model = svm.SVC() model.fit(train_X, train_y) prediction = model.predict(test_X) print('The accuracy of the SVM is: {0}'.format(metrics.accuracy_score(prediction,test_y))) The accuracy of the SVM is: 1.0 # Logistic Regression model = LogisticRegression() model.fit(train_X, train_y) prediction = model.predict(test_X) print('The accuracy of the Logistic Regression is: {0}'.format(metrics.accuracy_score(prediction,test_y))) The accuracy of the Logistic Regression is: 0.9555555555555556 # Decision Tree model=DecisionTreeClassifier() model.fit(train_X, train_y) prediction = model.predict(test_X) print('The accuracy of the Decision Tree is: {0}'.format(metrics.accuracy_score(prediction,test_y))) The accuracy of the Decision Tree is: 0.9555555555555556 # K-Nearest Neighbours model=KNeighborsClassifier(n_neighbors=3) model.fit(train_X, train_y) prediction = model.predict(test_X) print('The accuracy of the KNN is: {0}'.format(metrics.accuracy_score(prediction,test_y))) The accuracy of the KNN is: 1.0 上面使用了數據集的所有特征，下面將分別使用花瓣和花萼的尺寸：
petal = iris[['PetalLengthCm', 'PetalWidthCm', 'Species']] train_p,test_p=train_test_split(petal,test_size=0.3,random_state=0) train_x_p=train_p[['PetalWidthCm','PetalLengthCm']] train_y_p=train_p.Species test_x_p=test_p[['PetalWidthCm','PetalLengthCm']] test_y_p=test_p.Species sepal = iris[['SepalLengthCm', 'SepalWidthCm', 'Species']] train_s,test_s=train_test_split(sepal,test_size=0.3,random_state=0) train_x_s=train_s[['SepalWidthCm','SepalLengthCm']] train_y_s=train_s.Species test_x_s=test_s[['SepalWidthCm','SepalLengthCm']] test_y_s=test_s.Species model=svm.SVC()
model.fit(train_x_p,train_y_p)
prediction=model.predict(test_x_p)
print('The accuracy of the SVM using Petals is: {0}'.format(metrics.accuracy_score(prediction,test_y_p))) model.fit(train_x_s,train_y_s) prediction=model.predict(test_x_s) print('The accuracy of the SVM using Sepal is: {0}'.format(metrics.accuracy_score(prediction,test_y_s))) The accuracy of the SVM using Petals is: 0.9777777777777777 The accuracy of the SVM using Sepal is: 0.8 model = LogisticRegression()
model.fit(train_x_p, train_y_p)
prediction = model.predict(test_x_p)
print('The accuracy of the Logistic Regression using Petals is: {0}'.format(metrics.accuracy_score(prediction,test_y_p))) model.fit(train_x_s, train_y_s) prediction = model.predict(test_x_s) print('The accuracy of the Logistic Regression using Sepals is: {0}'.format(metrics.accuracy_score(prediction,test_y_s))) The accuracy of the Logistic Regression using Petals is: 0.6888888888888889 The accuracy of the Logistic Regression using Sepals is: 0.6444444444444445 model=DecisionTreeClassifier()
model.fit(train_x_p, train_y_p)
prediction = model.predict(test_x_p)
print('The accuracy of the Decision Tree using Petals is: {0}'.format(metrics.accuracy_score(prediction,test_y_p))) model.fit(train_x_s, train_y_s) prediction = model.predict(test_x_s) print('The accuracy of the Decision Tree using Sepals is: {0}'.format(metrics.accuracy_score(prediction,test_y_s))) The accuracy of the Decision Tree using Petals is: 0.9555555555555556 The accuracy of the Decision Tree using Sepals is: 0.6666666666666666 model=KNeighborsClassifier(n_neighbors=3) model.fit(train_x_p, train_y_p) prediction = model.predict(test_x_p) print('The accuracy of the KNN using Petals is: {0}'.format(metrics.accuracy_score(prediction,test_y_p))) model.fit(train_x_s, train_y_s) prediction = model.predict(test_x_s) print('The accuracy of the KNN using Sepals is: {0}'.format(metrics.accuracy_score(prediction,test_y_s))) The accuracy of the KNN using Petals is: 0.9777777777777777 The accuracy of the KNN using Sepals is: 0.7333333333333333 從中不難看出，使用花瓣的尺寸來訓練數據較花萼更准確。正如在探索性分析的熱圖中所看到的那樣，花萼的寬度和長度之間的相關性非常低，而花瓣的寬度和長度之間的相關性非常高。