Data can be downloaded as a zipped archive (FallData.tar.gz, ~330MB):

• link 1

Alternatively, running in a terminal

python download_data.py

automatically downloads and extracts the data. This code requires Python 3 and tqdm (pip install tqdm).

Once extracted, the data can be read using the following code snippets (in Python, R). Be sure to execute those lines while in the same directory as the extracted FallData folder.

Signals are loaded into Numpy arrays. Please be sure to have it installed (pip install numpy), as well as tqdm (pip install tqdm).

from load_data import get_code_list, load_signal, load_metadata


# Load and manipulate all signals and metadata.
all_codes = get_code_list()
print("There are {} trials.".format(len(all_codes)))
for code in all_codes:
    signal = load_signal(code)  # numpy array (n_samples, n_dims)
    metadata = load_metadata(code)  # dictionary
    # Do something.
    # ...

Be sure to set the working directory (with the function setwd) to wherever the data file has been unzipped. To read JSON files, the package jsonlite must be installed.

library("jsonlite")
code_list <- fromJSON("code_list.json")

for(code in code_list){
    if (startsWith(code, "u-")){
        filename <- paste("FallData/Unconstrained/", code, sep="")
    }
    if (startsWith(code, "c-")){
        filename <- paste("FallData/Controlled/", code, sep="")
    }
    if (startsWith(code, "cnf-")){
        filename <- paste("FallData/ControlledNoFall/", code, sep="")
    }
    signal <- read.csv(paste(filename, ".csv", sep=""))
    metadata <- fromJSON(paste(filename, ".json", sep=""))
    # Do something.
    # ...
}