使用 GridSearchCV 和 OneClassSVM 时出现“NU”值错误

Question

我正在使用 GridSearchCV 为我的模型 OneClassSVM 寻找最佳参数。我在文档中读到 nu 值应该在 0 到 1 之间。在我的代码中，我总是出错：

ValueError: nu <= 0 or nu > 1

我做错了什么？

import pandas as pd
from sklearn.svm import OneClassSVM

from sklearn.model_selection import GridSearchCV

import numpy as np

df = pd.read_csv('pitanja.csv')
#print(df)

x = df.iloc[:,:-1]

tuned_svm = {'cache_size':[70,80,90], 'coef0':[0.25,0.5,0.75], 'gamma':['auto'],
             'kernel':['poly','rbf','linear','sigmoid', 'precomputed'], 'random_state':[None],
             'shrinking':[True,False], 'tol':[0.05,0.1,0.2], 'verbose':[True],
             'nu':[0.1, 0.11, 0.12, 0.13, 0.14, 0.15, 0.16, 0.17, 0.18, 0.19, 0.2, 0.21, 0.22, 0.23, 0.24,0,25,0.26,0.27]}

def scorer_f(estimator, X):   #your own scorer
      return np.mean(estimator.score_samples(X))

out_cls = GridSearchCV(OneClassSVM(), tuned_svm, scorer_f)          

r = df

model = out_cls .fit(x)
prediction = model.predict(x)

print(model.best_params_)

Answer 1

发现错误：

'nu':[0.1, 0.11, 0.12, 0.13, 0.14, 0.15, 0.16, 0.17, 0.18, 0.19, 0.2, 0.21, 0.22, 0.23, 0.24,0,25 ,0.26,0.27]

如果您厌倦了手动输入这些值，我建议使用 'nu':[0.1, np.linspace(0.11,0.27,17)]。

使用 GridSearchCV 和 OneClassSVM 时出现“NU”值错误

Error whit `NU` value when using GridSearchCV and OneClassSVM

python

outliers

scikit-learn