AWS_ML

This repository contains two data science and machine learning projects that demonstrate applied modeling, experimentation, and notebook-driven analysis.

Projects

1. Handwritten Digits Classifier with PyTorch

Directory: Handwritten_Digits_Classifier_with_PyTorch

A PyTorch-based project for handwritten digit classification using the MNIST dataset.

• Goal: Train and evaluate a neural network to recognize handwritten digits (0-9).
• Framework: PyTorch
• Key Topics: data loading, preprocessing, model definition, training, evaluation, visualization, and model persistence.
• Files:
  • MNIST_Handwritten_Digits-STARTER.ipynb — notebook with the full training and evaluation workflow.
  • HandwrittenDigitsClassifier.pth — saved model weights.
  • requirements.txt — Python dependencies for running the notebook and code.

2. Predict Bike Sharing Demand with AutoGluon

Directory: Predict_Bike_Sharing_Demand_with_AutoGluon

A Kaggle-style regression project using AutoGluon to predict bike rental demand.

• Goal: Build and improve a model to predict bike sharing demand from weather and time features.
• Framework: AutoGluon (Tabular Prediction)
• Key Topics: exploratory data analysis, feature engineering, automated modeling, hyperparameter tuning, and competition submission.
• Files:
  • project-template.ipynb and project-template_.ipynb — notebooks for the project workflow.
  • train.csv, test.csv, sampleSubmission.csv — Kaggle dataset and template submission files.
  • submission.csv, submission_new_features.csv, submission_new_hpo.csv — example model output submissions.
  • model_test_score.png — sample model output or evaluation visualization.
  • report-template.md — report template for documenting experiments.
  • LICENSE.txt — license information for the project folder.

How to Use This Repository

1. Open the folder for the project you want to explore.
2. Install the dependencies listed in that folder's requirements.txt or follow the project-specific instructions inside its README.md.
3. Launch Jupyter or JupyterLab and run the notebook(s) to reproduce the analysis.
4. For the bike sharing project, the Kaggle dataset files and example submissions are included in the folder.

Notes

• The root repository does not include a single consolidated license for all projects. Use the folder-specific LICENSE.txt file in Predict_Bike_Sharing_Demand_with_AutoGluon or apply your own licensing terms.
• Each project is self-contained in its respective directory.