Skip to content

Latest commit

 

History

History
46 lines (28 loc) · 2.83 KB

README.md

File metadata and controls

46 lines (28 loc) · 2.83 KB

Activity Classification

Open In Colab

This project involves processing and analyzing data from the accelerometer and gyroscope sensors of an Apple Watch to classify various activities.

Purpose

The purpose of this project is to develop a model that can classify various activities based on data from the accelerometer and gyroscope sensors of an Apple Watch. This can potentially have a variety of applications, such as tracking physical activity or monitoring the movements of individuals for safety purposes.

Code

The code for processing and analyzing the data is written in Python and makes use of the pandas library. The code includes several helper functions for handling timestamps and splitting lists into smaller chunks. To run the code, you will need to have the following dependencies installed:

pandas

Instructions for running the code and any necessary input/output files are included in the comments of the code itself. The code performs the following tasks:

* Import necessary libraries and constants
* Read in .csv files containing the raw sensor data from multiple sessions
* Load a .json file containing activity labels for each session
* Process and clean the sensor data, including converting timestamps to a consistent format and selecting relevant columns
* Create a dataframe of activity labels and sensor data for each session
* Concatenate the dataframes for each session into a single dataframe
* Save the resulting dataframe to a .csv file

The resulting dataframe is saved to a .csv file in the output directory. The file is then organized into two sets: a training set and a testing set. Each set is further organized into sets by activity.

Data

The raw data from the accelerometer and gyroscope sensors is included in the data directory. The data is provided in .csv format and is organized by round of data collection (two rounds in total) and session within each round. Each session includes data from one user.

Model Training and Testing

The cleaned activity data was used to train and test the classification model. The model was implemented using CoreML.

References

Hutcherson, T. (2019, October 9). Activity classification with create ML, COREML3, and skafos: Part 1. Medium. Retrieved January 2, 2023, from https://medium.com/@tyler.hutcherson/activity-classification-with-create-ml-coreml3-and-skafos-part-1-8f130b5701f6