查看特徵的重要程度(以RF為例)

[ML100-Day22]Feature Importance.py

from sklearn.ensemble import RandomForestClassifier
import matplotlib.pyplot as plt

feature_labels = list(train_X.columns) # 欄位名稱存下來等下顯示用
forest = RandomForestClassifier().fit(train_X, train_Y) # 送進模型fit

# 用`.feature_importances_`取得重要性
importances = forest.feature_importances_ 
# 取得對應的index，等下顯示用
indices = np.argsort(importances)[::-1]

# 印出所有重要性特徵
for f in range(train_X.shape[1]):
    print("%2d) %-*s %f" % (f + 1, 30, 
                            feature_labels[indices[f]], 
                            importances[indices[f]]))
# 圖形顯示
plt.title('Feature Importance')
plt.bar(range(train_X.shape[1]), 
        importances[indices],
        align='center')

plt.xticks(range(train_X.shape[1]), 
           [feature_labels[x] for x in indices], rotation=45)
plt.xlim([-1, train_X.shape[1]])
plt.tight_layout()
plt.show()

欄位名稱：

重要程度：

print出來顯示

圖形顯示