def preprocess_data(filename='/content/drive/My Drive/fer2013.csv',
image_size=(48, 48)):
X, Y = getData(filename)
num_class = len(set(Y))
# balance = balance_class(Y)
N, D = X.shape
X = X.reshape(N, image_size, 1)
return give_train_test_splits(X, Y, test_size=0.1,
random_state=0), num_class
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()
from sklearn.model_selection import GridSearchCV
import pandas as pd
import warnings
import matplotlib.pyplot as plt
import statsmodels.api as sm
from sklearn.svm import SVR
from dataset import getData, plot, RMSPE, RMSE
import numpy as np
import time
start = time.clock()
warnings.filterwarnings('ignore')
X_train, Y_train, X_test, Y_test = getData()
#X_train.shape (882, 6)
#Y_train.shape (882,)
print(X_train)
print(Y_train)
'''
svr = GridSearchCV(SVR(), param_grid={"kernel": ("linear", 'rbf'), "C": np.logspace(-3, 3, 7), "gamma": np.logspace(-3, 3, 7)})
svr.fit(X_train, Y_train)
y_predict = svr.predict(X_test)
open = X_test['Open'].values
for i in range(len(open)):
if open[i] == 0:
#rmse = np.sqrt(mean_squared_error(test_predict,test_y))
#mae = mean_absolute_error(y_pred=test_predict, y_true=test_y)
#print('mae:', mae, ' rmse:', rmse)
return test_predict,xx,yy
test_predict,xx,yy = train_lstm(batch_size=80,time_step=15,train_begin=0,train_end=882)
predict = []
for i in range(4):
for j in test_predict[i]:
predict.append(j)
X_train1,Y_train1,X_test1,Y_test1 = getData()
open = X_test1['Open'].values
for i in range(len(open)):
if open[i] == 0:
predict[i] = 0
print(predict)
predict = pd.DataFrame({'Date':pd.date_range(start='2015-06-02',end='2015-07-31'),'Sales':predict})
predict = predict.set_index('Date')
rmspe = RMSPE(Y_test1.values,predict.values)
print(rmspe)
rmse = RMSE(Y_test1.values,predict.values)
print(rmse)
