近红外光谱分析技术属于交叉领域,需要化学、计算机科学、生物科学等多领域的合作。为此,在(北邮邮电大学杨辉华老师团队)指导下,近期准备开源传统的PLS,SVM,ANN,RF等经典算和SG,MSC,一阶导,二阶导等预处理以及GA等波长选择算法以及CNN、AE等最新深度学习算法,以帮助其他专业的更容易建立具有良好预测能力和鲁棒性的近红外光谱模型。
本篇主要讲述基于基于PCA和PLS的近红外光谱建模方法,
#载入数据
data_path = './/data//m5.csv' #数据
label_path = './/data//label.csv' #标签(反射率)
data = np.loadtxt(open(data_path, 'rb'), dtype=np.float64, delimiter=',', skiprows=0)
label = np.loadtxt(open(label_path, 'rb'), dtype=np.float64, delimiter=',', skiprows=0)
# 绘制原始后图片
plt.figure(500)
x_col = np.linspace(0,len(data[0,:]),len(data[0,:])) #数组逆序
y_col = np.transpose(data)
plt.plot(x_col, y_col)
plt.xlabel("Wavenumber(nm)")
plt.ylabel("Absorbance")
plt.title("The spectrum of the corn dataset",fontweight= "semibold",fontsize='x-large')
plt.savefig('.//Result//MSC.png')
plt.show()
#随机划分数据集
x_data = np.array(data)
y_data = np.array(label[:,2])
test_ratio = 0.2
X_train,X_test,y_train,y_test = train_test_split(x_data,y_data,test_size=test_ratio,shuffle=True,random_state=2)
#载入数据
#PCA降维到10个维度,测试该数据最好
pca=PCA(n_components=10) #只保留2个特征
pca.fit(X_train)
X_train_reduction = pca.transform(X_train)
X_test_reduction = pca.transform(X_test)
# PCA降维后图片绘制
plt.figure(100)
plt.scatter(X_train_reduction[:, 0], X_train_reduction[:, 1],marker='o')
plt.xlabel("Wavenumber(nm)")
plt.ylabel("Absorbance")
plt.title("The PCA for corn dataset",fontweight= "semibold",fontsize='large')
plt.savefig('.//Result//PCA.png')
plt.show()
#pls预测
pls2 = PLSRegression(n_components=3)
pls2.fit(X_train_reduction, y_train)
train_pred = pls2.predict(X_train_reduction)
pred = pls2.predict(X_test_reduction)
#计算R2
train_R2 = r2_score(train_pred,y_train)
R2 = r2_score(y_test,pred) #Y_true, Pred
print('训练R2:{}'.format(train_R2))
print('测试R2:{}'.format(R2))
#计算MSE
print('********************')
x_MSE = mean_squared_error(train_pred,y_train)
t_MSE = mean_squared_error(y_test,pred)
print('训练MSE:{}'.format(x_MSE))
print('测试MSE:{}'.format(t_MSE))
#计算RMSE
print('********************')
print('测试RMSE:{}'.format(sqrt(x_MSE)))
print('训练RMSE:{}'.format(sqrt(t_MSE)))
#绘制拟合图片
plt.figure(figsize=(6,4))
x_col = np.linspace(0,16,16) #数组逆序
# y = [0,10,20,30,40,50,60,70,80]
# x_col = X_test
y_test = np.transpose(y_test)
ax = plt.gca()
ax.set_xlim(0,16)
ax.set_ylim(6,11)
# plt.yticks(y)
plt.scatter(x_col, y_test,label='Ture', color='blue')
plt.plot(x_col, pred,label='predict', marker='D',color='red')
plt.legend(loc='best')
plt.xlabel("测试集的样本")
plt.ylabel("样本的值")
plt.title("The Result of corn dataset",fontweight= "semibold",fontsize='large')
plt.savefig('.//Result//Reslut.png')
plt.show()
完整代码可从获得GitHub仓库
代码仅供学术使用,如需问题,联系方式:QQ:1427950662,微信:Fu_siry