import torch
import numpy as np
import pandas as pd
import torch.nn as nn

# Defining Dummy Dataset and Parameters for a Mdel
X = torch.tensor([[1],[2],[3],[4],[5]], dtype = torch.float32)
y = torch.tensor([[2],[4],[6],[8],[10]], dtype = torch.float32)

# Extracting n_rows and n_cols for a simple ANN
n_rows, n_cols = X.shape

# Creating a Test Tensor to check the prediction before any Training
test = torch.tensor([5.0], dtype = torch.float32)

# 1st Method : By Using [General Class] for defining Linear Regression Model


# Linear_Regression Class containing functions like :
    # 1- (layers initialization function)
    # 2- (Forward function) 

class Linear_Regression(nn.Module):
    
    # __init__ Function of a Class 
    def __init__(self, input_size, output_size):
        
        super(Linear_Regression, self).__init__()
        
        # Defining the layers of the Simple ANN
        self.layer1 = nn.Linear(input_size, output_size)
    
    # Forward Function for a Layers defined in __init()__ Function
    def forward(self, X):
        return self.layer1(X)

input_size = n_cols
output_size = n_cols
model = Linear_Regression(input_size, output_size)

# Prediction a Test Tensor before any Training process
print(f'Prediction for f(5) is : {model.forward(test).item():.3f}')

# 2nd Method : By Using [PyTorch nn.Linear()] function to define Linear Regression Model


# input_size = n_cols
# output_size = n_cols
# model = nn.Linear(input_size, output_size)

# Defining Learning Rate for the Model
learning_rate = 0.02

# Loss Function
loss = nn.MSELoss()

# Optimizer Function
optimizer = torch.optim.SGD(model.parameters(), lr = learning_rate)