DATA1030 Final Project Dominique Barnes. Created a pipeline for a machine learning model to predict physical activity class on new unseen users. The data was collected by the WIreless Senor Data Mining Lab (WISDM) at Forham University in 2012. The goal of the final project was to develop a model classifier to identify six physical activity classes (downstairs, jogging, sitting, standing, walking, upstairs) through phone-based accelerometer readings. The dataset was source from Kaggle.
The project was built on Python 3.11.4 and repository organization as follows
- data/ - Stores all raw and preprocessed data files from the WISDM Lab
- figures/ - Stores ll visualizations including EDA and model result comparions
- results/ - Trained models pickled files
- report/ - Reports on development, pipeline, methodology, and model results. Presented in word document as final report and powerpoint presentation
- src/ - All of the code for: data importing, EDA; group based splitting, preprocessing and model development; model evaluations, result visualizations and feature importances in IMU_Final_Project.ipynb
- Key packages used in this project are the following:
- 'numpy': "1.24.4"
- 'matplotlib': "3.7.2"
- 'sklearn': "1.3.0"
- 'pandas': "2.0.3"
- 'xgboost': "1.7.6"
- 'shap': "0.42.1"
- 'seaborn': "0.12.2"