This project demonstrates how to build a binary classification model for tabular data using PyTorch. The workflow is implemented in the notebook Tabular_Data_Classification.ipynb and uses a rice type classification dataset from Kaggle.
- Dataset Download: Automatically downloads the rice type classification dataset using
opendatasets. - Data Preprocessing: Cleans the data by removing missing values and normalizes features for better model performance.
- Data Splitting: Splits the data into training, validation, and test sets.
- PyTorch Dataset & DataLoader: Wraps the data in custom PyTorch Dataset objects and uses DataLoader for efficient batching.
- Model Definition: Implements a simple feedforward neural network for binary classification.
- Training & Validation: Trains the model and tracks loss and accuracy for both training and validation sets.
- Testing: Evaluates the model on the test set and reports accuracy.
- Visualization: Plots training/validation loss and accuracy over epochs.
- Inference: Allows user input for feature values to make predictions with the trained model.
- Open
Tabular_Data_Classification.ipynbin Jupyter or VS Code. - Run all cells sequentially to:
- Download and preprocess the dataset
- Train and evaluate the model
- Visualize results
- Make predictions using custom input
- Python 3.7+
- PyTorch
- scikit-learn
- matplotlib
- pandas
- numpy
- opendatasets
- torchsummary
Install dependencies with:
!pip install opendatasets torch torchsummary scikit-learn matplotlib pandas numpy- The notebook is designed for educational purposes and can be adapted for other tabular classification tasks.
- GPU acceleration is supported if available.
This project is provided for educational use. Please check the dataset license before using for commercial purposes.