LSTM 预测各国人均GPD(Pytorch)

LSTM长短期记忆神经网络的学习与实现

「公开课干货分享」LSTM长短期记忆神经网络的学习与实现

 

 

以下是训练过程：

from pandas_datareader.pandas_datareader import wb
import torch.nn
import torch
import torch.optim
import csv
from IPython.display import display
import pandas as pd 
import numpy
import matplotlib.pyplot as plt
 
class Net(torch.nn.Module):
    def __init__(self,input_size,hidden_size):
        super(Net,self).__init__()
        self.rnn = torch.nn.LSTM(input_size,hidden_size)
        self.fc = torch.nn.Linear(hidden_size,1)
 
    def forward(self, x):
        x = x[:,:,None]
        x, _ = self.rnn(x)
        x = self.fc(x)
        x = x[:,:,0]
        return x
 
countries=['BR', 'CA', 'CN', 'FR', 'DE', 'IN', 'IL','JP', 'SA', 'GB', 'US']
dat = wb.download(indicator='NY.GDP.PCAP.KD',country=countries,start=1970,end=2016)
df = dat.unstack().T
df.index = df.index.droplevel(0)
year_num, sample_num = df.shape
 
countries = df.columns
 
years = df.index
# with open('F:/tracker_programe/lstm_test/1970_2016.csv', 'w') as f:
#     writer = csv.writer(f)
#     for row in df:
#         for col in row:
#             writer.writerow(col)
# print(df)
 
net = Net(input_size=1, hidden_size=5)
 
print(net)
 
df_scaled = df / df.loc['2000']
 
years = df.index
train_seq_len = sum((years >= '1971')&(years <= '2000'))
test_seq_len = sum(years > '2000')
 
print('TrainSet Length = {:.0f}, TestSet Length ={:.0f}'.format(train_seq_len, test_seq_len))
 
inputs=torch.tensor(df_scaled.iloc[:-1].values,dtype=torch.float32)
labels=torch.tensor(df_scaled.iloc[1:].values,dtype=torch.float32)
 
criterion = torch.nn.MSELoss()
optmizer = torch.optim.Adam(net.parameters())
for step in range(10001):
    if step:
        optmizer.zero_grad()
        train_loss.backward()
        optmizer.step()
    preds = net(inputs)
    train_preds = preds[:train_seq_len]
    train_labels = labels[:train_seq_len]
    train_loss = criterion(train_preds,train_labels)
 
    test_preds = preds[-test_seq_len]
    test_labels = labels[-test_seq_len]
    test_loss = criterion(test_preds,test_labels)
 
    if step % 500 ==0:
        print('Run {:.0f}: loss(TrainSet)={:.6f}, loss(TestSet)={:.6f}'.format(step, train_loss, test_loss))
 
preds=net(inputs)
df_pred_scaled=pd.DataFrame(preds.detach().numpy(), index=years[1:], columns=df.columns)
df_pred=df_pred_scaled*df.loc['2000']
display((df_pred.loc['2001':]-df['2001':])/df['2001':])
 
# Code borrow from：凌空的桨
# https://blog.csdn.net/baidu_36669549/article/details/85267766