Activity Recognition from Accelerometer Data

This repository contains R code to apply the activity classification models presented in the paper "Accelerometer-Based Classification of Walking, Running, and Jumping: A Machine Learning Approach for Mechanical Loading Assessment".

Prerequisites

Before using this code, you need to:

1. Install R (version 4.1.0 or higher) from https://cran.r-project.org/
2. Install RStudio from https://posit.co/download/rstudio-desktop/

Setup

1. Download this project (click the green "Code" button and select "Download ZIP")
2. Unzip the downloaded file
3. Open the activity_recognition.Rproj file. This should open the project in RStudio
4. In the RStudio Console, run:

install.packages("renv")
renv::restore()

This will install all required packages.

Using the code

Run the analysis

1. Open main.R in RStudio
2. Update these lines with your data details:

data <- vroom(
  "path/to/your/data",  # Replace with path to your CSV file
  skip = 10             # Replace with number of header lines (0 if none)
)

results <- classify_activities(
  data = data,
  time_col = "timestamp",  # Replace with your timestamp column name
  x_col = "acc_x",         # Replace with your X acceleration column name
  y_col = "acc_y",         # Replace with your Y acceleration column name
  z_col = "acc_z",         # Replace with your Z acceleration column name
  sampling_freq = 100,     # Replace with your sampling frequency in Hz
  placement = "ankle",     # Choose: "ankle", "lower_back", or "hip"
  model_type = "rf"        # Choose: "rf", "svm", or "knn"
)

3. Click the "Source" button (or press Ctrl+Shift+S) to run the analysis

Common issues

1. "File not found" error: Make sure the path to your CSV file is correct
2. Column name errors: Check that your column names match exactly what you specified
3. Package loading errors: Run renv::restore() again to ensure all packages are installed

Need Help?

If you find any bugs or have questions, requests, or comments, please report them in this repository's issues page