1.测试集准确率的计算

# 保存模型
PATH = './NIR.pth'
torch.save(network.state_dict(), PATH)
# 加载模型
network = Network()
network.load_state_dict(torch.load(PATH))

@torch.no_grad()
def get_all_preds(model, loader):
    all_preds = torch.tensor([])
    print(len(loader))
    for batch in loader:
        images, labels = batch

        preds = model(images)
        all_preds = torch.cat((all_preds, preds) ,dim=0)

    return all_preds

test_preds = get_all_preds(network, test_loader)
actual_labels_test = torch.Tensor(test_data.labels)
preds_correct_test = test_preds.argmax(dim=1).eq(actual_labels_test).sum().item()

print('total correct:', preds_correct_test)
print('accuracy_test:', preds_correct_test / len(test_data))

这里注意一个问题，我想知道我在训练集上的模型准确率，直接按照accuracy_test的计算方式进行计算，我适用的是train_data.laber。但是这里我发现，因为我设置了batch_size为16，而数据总数是415，所以我发现在经过batch_size后，我的train_loader为25个batch_size，即为400个，train_data为415，这个就是在batch_size无法整除数据时所造成的问题。

那么从另一个点上也就是说，我实际上只有400个数据参与了模型训练，而不是415个。

2.输出最后一层的预测概率

调整输出的灵敏度和特异度的时候需要调整预测概率。验证集的训练方法如下：

@torch.no_grad()
def get_all_preds(model, loader):
    all_preds = torch.tensor([])
    for batch in loader:
        images, labels = batch
        
        preds = model(images)
        all_preds = torch.cat((all_preds, preds) ,dim=0)

    return all_preds

test_preds = get_all_preds(network, test_loader)
actual_labels_test = torch.Tensor(test_data.labels)
preds_correct_test_list = []
#preds_correct_test = torch.softmax(test_preds,dim=0)[1][0].cpu().item()
for i in range(len(test_preds)):
#     preds_correct_test_list.append(torch.softmax(test_preds,dim=1)[i][0].cpu().item())
    preds_correct_test_list.append(torch.softmax(test_preds,dim=1)[i][1].cpu().item())

3.手动计算灵敏度，特异度，准确率等参数

# 手动计算敏感度特异度
sum_number_0 = 0
sum_number_1 = 0
for i in range(len(preds_correct_test_list)):
    # 输入调整的概率
    if preds_correct_test_list[i] > 0.00000000005:
        preds_correct_test_list[i] = 1 
    else:
        preds_correct_test_list[i] = 0
        
TP,TN,FP,FN = 0,0,0,0
for i in range(len(test_loader.dataset.labels)):
    if (test_loader.dataset.labels[i]==0) and (preds_correct_test_list[i]==0):
        TP += 1 
    if (test_loader.dataset.labels[i]==0) and (preds_correct_test_list[i]==1):
        FN += 1 
    if (test_loader.dataset.labels[i]==1) and (preds_correct_test_list[i]==0):
        FP += 1 
    if (test_loader.dataset.labels[i]==1) and (preds_correct_test_list[i]==1):
        TN += 1 

accuracy = (TP+TN)/(TP+TN+FP+FN)
precision = TP/(TP+FP)
# 敏感度特异度
sensitivity = TP/(TP+FN)
specificity = TN/(TN+FP)
print(accuracy,precision,sensitivity,specificity)

4.画一个matrix图，展示正确时错误的数据

# 画一个matrix图展示正确和错误的数据
import itertools
import numpy as np
import matplotlib.pyplot as plt

def plot_confusion_matrix(cm, classes,
                          normalize=False,
                          title='Confusion matrix',
                          cmap=plt.cm.Blues):
    """
    This function prints and plots the confusion matrix.
    Normalization can be applied by setting `normalize=True`.
    """
    if normalize:
        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]
        print("Normalized confusion matrix")
    else:
        print('Confusion matrix, without normalization')

    print(cm)

    plt.imshow(cm, interpolation='nearest', cmap=cmap)
    plt.title(title)
    plt.colorbar()
    tick_marks = np.arange(len(classes))
    plt.xticks(tick_marks, classes, rotation=45)
    plt.yticks(tick_marks, classes)

    fmt = '.2f' if normalize else 'd'
    thresh = cm.max() / 2.
    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):
        plt.text(j, i, format(cm[i, j], fmt),
                 horizontalalignment="center",
                 color="white" if cm[i, j] > thresh else "black")

    plt.tight_layout()
    plt.ylabel('True label')
    plt.xlabel('Predicted label')

cm = confusion_matrix(test_data.labels, test_preds.argmax(dim=1))
classes = ('0','1')
plt.figure(figsize=(10,10))
plot_confusion_matrix(cm, classes)

5.画LOSS曲线图

最简单的一种方法

1
2
3

plt.plot(loss_list)
plt.legend()
plt.title('Compare loss for different models in training')

将LOSS和ACC画到一起的

from mpl_toolkits.axes_grid1 import host_subplot
def plot_acc_loss(loss, acc):
    host = host_subplot(111)  # row=1 col=1 first pic
    plt.subplots_adjust(right=0.8)  # ajust the right boundary of the plot window
    par1 = host.twinx()   # 共享x轴
 
    # set labels
    host.set_xlabel("steps")
    host.set_ylabel("train-loss")
    par1.set_ylabel("train-accuracy")
 
    # plot curves
    p1, = host.plot(range(len(loss)), loss, label="loss")
    #p2, = par1.plot(range(len(acc)), acc, label="accuracy")
 
    # set location of the legend,
    # 1->rightup corner, 2->leftup corner, 3->leftdown corner
    # 4->rightdown corner, 5->rightmid ...
    host.legend(loc=5)
 
    # set label color
    host.axis["left"].label.set_color(p1.get_color())
    par1.axis["right"].label.set_color(p2.get_color())
 
    # set the range of x axis of host and y axis of par1
    # host.set_xlim([-200, 5200])
    # par1.set_ylim([-0.1, 1.1])
 
    plt.draw()
    plt.show()

plot_acc_loss(loss_list, acc_list)