37 lines (24 loc) · 1020 Bytes

Tabular Data Classification using PyTorch

This project builds a deep learning classifier using PyTorch to classify rice types based on physical attributes from a structured CSV dataset.

Project Highlights

Cleaned and normalized a rice classification dataset
Custom PyTorch Dataset class for scalable data loading
Deep learning model trained with train/validation/test splits
Performance tracked using accuracy and loss metrics
Final evaluation on unseen test data

Dataset

The dataset contains several numerical features for different rice grains, with the target being the type/class of rice.

Columns: Numerical attributes of rice grains
Target: Class column indicating rice type

Model Architecture

Fully connected feedforward neural network
Binary classification with BCEWithLogitsLoss

Results

Real-time accuracy and loss tracking over 10 epochs
Final test accuracy printed at the end of training

Libraries Used

PyTorch
Scikit-learn
Pandas
NumPy
Matplotlib