This assignment builds feedforward neural networks to classify human activities from sensor data. You'll experiment with different architectures (varying depth and width), compare performance against traditional machine learning approaches (PCA + Random Forest from Unit 7), and learn when neural networks justify their added complexity.
The project uses the Human Activity Recognition (HAR) dataset containing sensor readings from smartphones to classify 6 different human activities:
- Walking
- Walking Upstairs
- Walking Downstairs
- Sitting
- Standing
- Laying
The dataset contains 561 features derived from accelerometer and gyroscope sensor data.
Assignment-6/
├── README.md # Project documentation
├── starter_notebook.ipynb # Main Jupyter notebook with assignments
└── data/
├── train.csv # Training dataset (sensor features + activity labels)
└── test.csv # Test dataset (sensor features + activity labels)
- Python 3.x
- Jupyter Notebook or VS Code with Jupyter extension
pandas
numpy
matplotlib
seaborn
scikit-learn
tensorflow
First, install the required packages. You can run this from the first code cell in the notebook:
pip install pandas numpy matplotlib seaborn scikit-learn tensorflowOpen starter_notebook.ipynb in Jupyter Notebook or VS Code.
Work through the notebook sequentially from top to bottom. The notebook is structured as a guided assignment with TODO comments indicating where you need to write code.
After completing all steps and reflections:
git add .
git commit -m 'completed neural networks assignment'
git pushSubmit your GitHub repository link on the course platform.