如何解决优化 MLP 回归分数
我试图使用 sklearn 库中的 MLPRegressor 对 Langevin 函数进行曲线拟合,但即使使用在线 0.1 的测试大小,我也无法得到 R^2 得分高于 0.97 所以我想知道是否有人可以帮助我找到优化算法和拟合的方法。
import numpy as np
import matplotlib.pylab as plt
from sklearn.model_selection import train_test_split
from sklearn.neural_network import MLPRegressor
from sklearn import datasets
from sklearn import metrics
from sklearn.metrics import r2_score
import seaborn as sns
import numpy.random as rd
x_i = np.linspace(0.2,1,1000)
y_i = 1/np.tanh(x_i) - 1/x_i * 1.5
#plt.plot(x_i,y_i,"o")
X_treino,X_teste,y_treino,y_teste = train_test_split(x_i.reshape(-1,1),y_i.reshape(-1,test_size = 0.50)
model = MLPRegressor(hidden_layer_sizes = 200,activation = "relu",random_state = 1,max_iter = 200)
model.fit(X_treino,y_treino)
expected_y1 = y_teste
predicted_y1 = model.predict(X_teste)
print(metrics.r2_score(expected_y1,predicted_y1))
plt.figure(figsize = (16,9))
plt.plot(X_teste,y_teste,"o",color = "red",label = "Esperado")
plt.plot(X_teste,predicted_y1,color = "blue",label = "Previsto")
plt.legend(fontsize = "large")
版权声明:本文内容由互联网用户自发贡献,该文观点与技术仅代表作者本人。本站仅提供信息存储空间服务,不拥有所有权,不承担相关法律责任。如发现本站有涉嫌侵权/违法违规的内容, 请发送邮件至 dio@foxmail.com 举报,一经查实,本站将立刻删除。