CMPT 353 final project - Fall Detection

A Static Report with all the findings and precice details of techniques used is in the file Report.pdf

Required libraries

- All of the required libraries needed to run the project successfully are listed in the requirements.txt file. 
- Notably, these are the libraries you most likely need:
    - numpy
    - pandas
    - matplotlib
    - pykalman
    - scipy
    - statsmodels
    - seaborn
    - sklearn
    - tabulate
- The project should be run in Python 3.

Order of execution and results produced

Way 1.
    1) Run 'clean_save.ipynb'
        - Results produced: filtered files for each scenario
            - Location saved: 'Data/Cleaned/X/''
            - X is the scenario
            
    2) Run 'transformation.ipynb'
        - Results produces: one transformed file for each scenario
            - Location saved: 'Data/Transformed/X'
            - X is the scenario
        
    3) Run 'statistics.ipynb'
        - Results produced: graphs of multiple inferential and statistical tests
        - Images only displayed in the notebook, not saved
        
    4) Run 'machine_learning.ipynb'
        - Results produced: images for ROC Curve and Confusion matrix for each model we created
        - Images only displayed in the notebook, not saved
        - Best models saved in location: "Models"
    
    5) Run 'predict.ipynb'
        - Imports Saved models from "Models" 
        - Predicts and print results for never seen data in "Data/testData"
    
Way 2.   
    1) Run 'app.py'
        This will use already trained model and produce results after cleaning and transforming the data.
        - Input: takes in an input recording data file name from the command line
        - Results produced: prints out whether if it was a fall or not with the accuracy score
        - How to run it: 
            Linux   : python3 app.py filename.csv 
            Windows : python app.py filename.csv
        
    Note:
        - The input file should have similar format to files in Data/walkSit/walkSit1.csv
        - Simalar files could be generated using IOS app : “Sensor Data Recorder” - Nils Ackermann
        - Use Acceleration.csv generated from the application