①SVM
【高清png圖和wps導入的pos檔案在原文連結】
![](https://img.laitimes.com/img/__Qf2AjLwojIjJCLyojI0JCLicmbw5CZmBjNiNWM5Q2YiNDM5QmZ2UjZhJGZ4gTZ1YTO5YjN48CX0JXZ252bj91Ztl2Lc52YucWbp5GZzNmLn9Gbi1yZtl2Lc9CX6MHc0RHaiojIsJye.png)
②代碼部分【原書中因sklearn版本問題,導包部分已修改】
# 使用鸢尾花資料集 SVC分類,記住SVC不會輸出每個類别的機率from sklearn import datasetsfrom sklearn.svm import SVCiris = datasets.load_iris()X= iris["data"][:,(2,3)] y = iris["target"]setosa_or_versicolor=(y==0)|(y==1)X = X[setosa_or_versicolor]y = y[setosa_or_versicolor]svc_clf = SVC(kernel='linear',C=float('inf'))svc_clf.fit(X,y)# 可預測svc_clf.predict([[2.5,1.7]])
# 非線性SVM分類from sklearn.datasets import make_moonsfrom sklearn.pipeline import Pipelinefrom sklearn.preprocessing import PolynomialFeaturesX, y = make_moons(n_samples=100, noise=0.15, random_state=42)polynoimal_svm_clf = Pipeline(( ("poly_features",PolynomialFeatures(degree=3)), ("scaler",StandardScaler()), ("svm_clf",LinearSVC(C=10,loss = 'hinge'))))polynoimal_svm_clf.fit(X,y)
其餘代碼參見原文連結。
移入Jupyter預設路徑即可運作。
(提取碼:m6jb)