def train():
model = lstm(input_size=args.input_size,
hidden_size=args.hidden_size,
num_layers=args.layers,
output_size=1,
dropout=args.dropout,
batch_first=args.batch_first)
model.to(args.device)
# 是否需要继续训练
if args.useGPU:
checkpoint = torch.load(args.save_file)
else:
checkpoint = torch.load(args.save_file,
map_location=lambda storage, loc: storage)
model.load_state_dict(checkpoint['state_dict'])
criterion = nn.MSELoss() # 定义损失函数
# optimizer = torch.optim.Adam(model.parameters(), lr=args.lr) # Adam梯度下降 学习率=0.001
optimizer = torch.optim.RMSprop(model.parameters(), lr=args.lr)
close_max, close_min, train_loader, test_loader = getData(
args.corpusFile, args.sequence_length, args.batch_size)
for i in range(args.epochs):
total_loss = 0
for idx, (data, label) in enumerate(train_loader):
if args.useGPU:
data1 = data.squeeze(1).cuda()
pred = model(Variable(data1).cuda())
# print(pred.shape)
pred = pred[1, :, :]
label = label.unsqueeze(1).cuda()
# print(label.shape)
else:
data1 = data.squeeze(1)
pred = model(Variable(data1))
pred = pred[1, :, :]
label = label.unsqueeze(1)
loss = criterion(pred, label)
optimizer.zero_grad()
loss.backward()
optimizer.step()
total_loss += loss.item()
if i % args.saveepochs == 0:
# torch.save(model, args.save_file)
torch.save({'state_dict': model.state_dict()}, args.save_file)
print('第%d epoch，保存模型' % i)
print(total_loss)
# torch.save(model, args.save_file)
torch.save({'state_dict': model.state_dict()}, args.save_file)
def eval():
# model = torch.load(args.save_file)
model = lstm(input_size=args.input_size,
hidden_size=args.hidden_size,
num_layers=args.layers,
output_size=1)
model.to(args.device)
if args.useGPU:
checkpoint = torch.load(args.save_file)
else:
checkpoint = torch.load(args.save_file,
map_location=lambda storage, loc: storage)
model.load_state_dict(checkpoint['state_dict'])
preds = []
labels = []
close_max, close_min, train_loader, test_loader = getData(
args.corpusFile, args.sequence_length, args.batch_size)
# for idx, (x, label) in enumerate(test_loader):
for idx, (x, label) in enumerate(train_loader):
if args.useGPU:
x = x.squeeze(1).cuda() # batch_size,seq_len,input_size
else:
x = x.squeeze(1)
pred = model(x)
list = pred.data.squeeze(1).tolist()
preds.extend(list[-1])
labels.extend(label.tolist())
# print(preds)
count = 0
acc = 0
for i in range(len(preds) - 1):
# print('预测值是%.2f,真实值是%.2f' % (
# preds[i][0] * (close_max - close_min) + close_min, labels[i][0] * (close_max - close_min) + close_min))
# print("%.2f" % preds[i][0],end=", ")
if (preds[i][0] > 0):
print("预测会涨", end=", ")
else:
print("预测会跌", end=", ")
if (labels[i] == 1.0):
print("实际涨了")
else:
print("实际跌了")
if ((preds[i][0] >= 0 and labels[i] == 1.0)
or (preds[i][0] <= 0 and labels[i] == -1.0)):
acc += 1
count += 1
print(acc, count)
print("实际预测准确率{0}%".format(acc * 1.0 / count * 100))
def eval():
# model = torch.load(args.save_file)
model = lstm(input_size=args.input_size, hidden_size=args.hidden_size, num_layers=args.layers , output_size=1)
model.to(args.device)
checkpoint = torch.load(args.save_file)
model.load_state_dict(checkpoint['state_dict'])
preds = []
labels = []
close_max, close_min, train_loader, test_loader = getData(args.corpusFile, args.sequence_length, args.batch_size)
for idx, (x, label) in enumerate(test_loader):
if args.useGPU:
x = x.squeeze(1).cuda() # batch_size,seq_len,input_size
else:
x = x.squeeze(1)
pred = model(x)
list = pred.data.squeeze(1).tolist()
preds.extend(list[-1])
labels.extend(label.tolist())
for i in range(len(preds)):
print('预测值是%.2f,真实值是%.2f' % (
preds[i][0] * (close_max - close_min) + close_min, labels[i] * (close_max - close_min) + close_min))
def eval():
# model = torch.load(args.save_file)
model = lstm(input_size=args.input_size, hidden_size=args.hidden_size, num_layers=args.layers , output_size=1)
model.to(args.device)
checkpoint = torch.load(args.save_file)
model.load_state_dict(checkpoint['state_dict'])
criterion = nn.MSELoss() # 定义损失函数
optimizer = torch.optim.Adam(model.parameters(), lr=args.lr) # Adam梯度下降 学习率=0.001
preds = []
labels = []
close_max, close_min, train_loader, test_loader = getData(args.corpusFile, args.sequence_length, args.batch_size)
for idx, (x, label) in enumerate(test_loader):
if args.useGPU:
x = x.squeeze(1).cuda() # batch_size,seq_len,input_size
else:
x = x.squeeze(1)
pred = model(Variable(x))
loss = criterion(pred, label)
optimizer.zero_grad()
loss.backward()
optimizer.step()
list = pred.data.squeeze(1).tolist()
preds.extend(list[-1])
labels.extend(label.tolist())
f = open('pred.csv','w',encoding='utf-8',newline='')
csv_writer = csv.writer(f)
for i in range(len(preds)):
print('预测值是%.2f,真实值是%.2f' % (
preds[i][0] * (close_max - close_min) + close_min, labels[i] * (close_max - close_min) + close_min))
csv_writer.writerow([preds[i][0],labels[i]])
f.close()