ShiroOnigami23/ames-house-price-predictor

Ames House Price Predictor

This repository hosts a PyTorch neural network model trained to predict house prices in Ames, Iowa, using the Ames 'house_prices' dataset.

Model Architecture

The model is a RegularizedModel which is a Multi-Layer Perceptron (MLP) with Dropout layers for regularization. The architecture is as follows:

RegularizedModel(
  (layer1): Linear(in_features=289, out_features=256, bias=True)
  (relu1): ReLU()
  (layer2): Linear(in_features=256, out_features=128, bias=True)
  (relu2): ReLU()
  (layer3): Linear(in_features=128, out_features=64, bias=True)
  (relu3): ReLU()
  (layer4): Linear(in_features=64, out_features=1, bias=True)
  (dropout1): Dropout(p=0.5, inplace=False)
  (dropout2): Dropout(p=0.5, inplace=False)
  (dropout3): Dropout(p=0.5, inplace=False)
)

Training Details

• Epochs: 2000
• Loss Function: Mean Squared Error (MSE)
• Optimizer: Adam (learning rate = 0.001)

Evaluation Metrics

The model was evaluated on a test set (20% of the data) after training.

• Mean Squared Error (MSE): 0.0019
• R-squared (R2): 0.8719

Usage (Example)

import torch
from huggingface_hub import hf_hub_download

# Assuming you have the RegularizedModel class defined as above
# input_features and output_features would be derived from your preprocessed data
input_features = 289 # Example value, replace with actual if different
output_features = 1

model = RegularizedModel(input_features, output_features)

# Download the model weights from Hugging Face Hub
model_path = hf_hub_download(repo_id="ShiroOnigami23/ames-house-price-predictor", filename="regularized_model.pth")
model.load_state_dict(torch.load(model_path))
model.eval()

# Example prediction (replace with your actual preprocessed input)
# dummy_input = torch.randn(1, input_features) # Ensure input is scaled like training data
# with torch.no_grad():
#    prediction = model(dummy_input)
# print(f"Predicted price (scaled): {prediction.item()}")

# To inverse transform the prediction to original price scale, you'd use your target_scaler
# original_price = target_scaler.inverse_transform(prediction.numpy())

Repository

• Model File: regularized_model.pth
• Repository ID: ShiroOnigami23/ames-house-price-predictor
• Hugging Face Profile: ShiroOnigami23