The goal of {trackballr} is to make analysis of trackball data easy-
You can install the development version of trackballr from GitHub with:
# install.packages("devtools")
devtools::install_github("roaldarbol/trackballr")or from my -universe:
install.packages("trackballr", repos = c("https://roaldarbol.r-universe.dev", "https://cran.r-project.org"))
This is a basic example which shows you how to solve a common problem:
library(trackballr)
## Read all files into list
data_list <- read_trackball_data(folder_path, sensors = c("Right", "Left"))
# Augment all data in list
data_list <- lapply(data_list, beetle_augment)
# Bind list into single data frame
data_df <- bind_rows(sensor_data_temp)Once the data has been pre-processed, it can then easily generate useful summaries - here are some examples.
# Compute translational summary
sensor_summary <- sensor_data_df |>
na.omit() |>
group_by(id, date) |>
filter(abs(x) > 0 | abs(y) > 0) |> # Only keep rows containing movement
summarise(total_translation = sum(distance),
v_translation_mean = mean(v_translation),
v_translation_sd = sd(v_translation),
sinuosity = sqrt(last(cum_x)^2 + last(cum_y)^2)/sum(distance)
) |>
mutate(trial = if_else(date <= min(date), "first", "second")) |> # Each animal has been on the trackball on two days - here we assign which day/trial
slice(1:2) |>
filter(total_translation > 0) |> # Filter away trials with no translation
filter(n() == 2) # Keep only observations where both trials are present
# Compute rotational summary
# Here wa are also filtering out trials with a faulty sensor/no data for one sensor
rotation_summary <- sensor_data_df |>
na.omit() |>
group_by(id, date) |>
filter(v_rotation > 0) |>
summarise(total_rotation = sum(rotation, na.rm = TRUE),
v_rotation_mean = mean(v_rotation, na.rm = TRUE)) |>
mutate(trial = if_else(date <= min(date), "first", "second")) |>
slice(1:2) |>
filter(n() == 2)