deep_learning_from_scratch_EXERCISE_1.py

deep_learning_from_scratch_EXERCISE_1.py

"""
Neural network prediction on a maintenance dataset using PyTorch.
Predicts whether a part needs replacement (1) or not (0).
Uses 3 nodes in the hidden layer and ReLU as the activation function.
"""

import pandas as pd
import torch
import torch.nn as nn
from torch.utils.data import DataLoader, TensorDataset

# Hyperparameters
LEARNING_RATE = 0.001
EPOCHS        = ?
BATCH_SIZE    = 32

# Data loading
df = pd.read_csv('https://bit.ly/3wlFsb4')

# Extract and scale input variables (all columns except last)
X = df.values[:, :-1] / 1000.0

# Extract output column (last column)
Y = df.values[:, -1]

# Convert to PyTorch tensors
X_tensor = torch.tensor(X, dtype=torch.float32)
Y_tensor = torch.tensor(Y, dtype=torch.float32).unsqueeze(1)  # shape: (N, 1)

dataset = TensorDataset(X_tensor, Y_tensor)
dataloader = DataLoader(dataset, batch_size=BATCH_SIZE, shuffle=True)

# Model declaration
n_features = X.shape[1]

model = nn.Sequential(
    nn.Linear(n_features, ?),   # hidden layer: n_features → 3 nodes
    ?,                  # ReLU activation
    nn.Linear(?, ?),            # output layer: 3 → 1 node
    ?               # Sigmoid for binary classification
)

# loss function and optimization
loss_fn   = nn.MSELoss()
optimizer = torch.optim.Adam(model.parameters(), lr=LEARNING_RATE)

# training
model.train()
for epoch in range(EPOCHS):
    epoch_loss = 0.0
    for X_batch, Y_batch in dataloader:
        optimizer.zero_grad()
        predictions = model(X_batch)
        loss = loss_fn(predictions, Y_batch)
        loss.backward()
        optimizer.step()
        epoch_loss += loss.item() * len(X_batch)

    if (epoch + 1) % 10 == 0:
        avg_loss = epoch_loss / len(dataset)
        print(f"Epoch {epoch + 1:>3}/{EPOCHS}  |  Loss: {avg_loss:.4f}")

# evaluate performance
# model.eval()
with torch.no_grad():
    all_preds  = model(X_tensor)
    # Round probabilities to 0 or 1 for accuracy calculation
    binary_preds = (all_preds >= 0.5).float()
    accuracy     = (binary_preds == Y_tensor).float().mean().item()

print(f"\nDataset Score (Accuracy): {accuracy:.4f}")