model.py

from sklearn.preprocessing import MinMaxScaler
import pandas as pd
import numpy as np
import torch
import matplotlib.pyplot as plt
df = pd.read_csv('apple_stock_data.csv')
vals = df.values
vals = vals[:,1:3]
print(vals)
scaler = MinMaxScaler()
scaled = scaler.fit_transform(vals)
print(scaled)
X = []
y = []
lookback = 100
for i in range(lookback, len(vals)):
    X.append(scaled[i-lookback:i])
    y.append(scaled[i][0])
y = np.array(y)
X = np.array(X)
print(X.shape, y.shape)
split_idx = int(len(X) * 0.8)
X_train = X[:split_idx]
y_train = y[:split_idx]
X_val = X[split_idx:]
y_val = y[split_idx:]
X_train= torch.tensor(X_train,dtype=torch.float32)
y_train= torch.tensor(y_train,dtype=torch.float32).view(-1,1)
X_val= torch.tensor(X_val,dtype=torch.float32)
y_val= torch.tensor(y_val,dtype=torch.float32).view(-1,1)
print(X_train.shape, y_train.shape)
print(X_val.shape, y_val.shape)
class LSTMModel(torch.nn.Module):
    def __init__(self):
        super(LSTMModel, self).__init__()
        self.input_dim = 2
        self.hidden_dim = 128
        self.layer_dim = 2
        self.output_dim = 1
        self.lstm = torch.nn.LSTM(self.input_dim, self.hidden_dim, self.layer_dim,dropout = 0.25,batch_first=True)
        self.linear = torch.nn.Linear(self.hidden_dim, self.output_dim)
    def forward(self,x):
        out,(hn,cn) = self.lstm(x)
        out = out[:,-1,:]
        out = self.linear(out)
        return out
model = LSTMModel()
criterion = torch.nn.MSELoss()
optimizer = torch.optim.Adam(model.parameters(), lr=0.001)
h0,c0 = None,None
for epoch in range(500):
    optimizer.zero_grad()
    model.train()
    outputs = model(X_train)
    loss = criterion(outputs,y_train)
    loss.backward()
    optimizer.step()
    if (epoch+1) % 5 == 0:
        print(f"epoch: {epoch+1}, loss: {loss.item():.4f}")
model.eval()
outputs= model(X_val).detach().numpy()
z = np.zeros((X_val.shape[0],2))
z[:,0] = outputs[:,0]
time_steps = np.arange(len(X_val))
outputs_inverted = scaler.inverse_transform(z)[:,0]
v = np.zeros((y_val.shape[0],2))
v[:,0] = y_val.detach().numpy().flatten()
y_val_inverse = scaler.inverse_transform(v)[:,0]
plt.plot(time_steps, outputs_inverted, time_steps,y_val_inverse)
plt.show()