为什么使用 sklearn 库对随机生成的数据的准确性低

Question

我生成了一个正态分布样本以及 3 个类来执行分类。我得到的准确度很低。我想知道您是否可以给我宝贵的反馈，以提高我的 LDA 分类器的性能。感谢您抽出宝贵时间。这是我的代码：

import pandas as pd
import numpy as np
from random import seed
import random
from sklearn.discriminant_analysis import LinearDiscriminantAnalysis
from sklearn.model_selection import cross_val_score
from sklearn.metrics import f1_score
from sklearn.metrics import accuracy_score
import time
seed(23)
mu, sigma = 0, 0.1 # mean and standard deviation
x1 = np.random.normal(mu, sigma, 1000)
x1=x1.reshape(-1, 1)
seed(1)
y=np.random.randint(0,3,size=(1000,1))
y_cross=np.ravel(y)
start_time1 = time.time()
clf_s=LinearDiscriminantAnalysis()
print('5-fold cross-validation accuracy score:', np.mean(cross_val_score(clf_s,x1, np.ravel(y), cv=5,scoring='accuracy')))
print('5-fold cross-validation F1 score:', np.mean(cross_val_score(clf_s, x1, np.ravel(y), cv=5,scoring='f1_micro')))
end_time1 = time.time()
print ("Computational time in seconds = " +str(end_time1 - start_time1) )

结果：

5-fold cross-validation accuracy score: 0.3280613765344133
5-fold cross-validation F1 score: 0.3280613765344133
Computational time in seconds = 1.4167194366455078

Answer 1

3个类的 .33 准确度意味着纯粹的猜测。我认为这是意料之中的，因为您生成的标签是随机的。算法应该发现数据中的结构。您准备数据的方式意味着您的算法没有可供学习的结构。如果您想要更高的准确性，请正确生成数据，例如使用sklearn.datasets.make_blobs并在该数据集上训练您的算法。

证明

from sklearn.discriminant_analysis import LinearDiscriminantAnalysis
from sklearn.model_selection import cross_val_score
from sklearn.datasets import make_blobs
X,y = make_blobs(n_samples=1000, n_features=2, centers=3, random_state=42)
clf=LinearDiscriminantAnalysis()
np.mean(cross_val_score(clf,X,y, cv=5,scoring='accuracy'))
1.0