v0.2 first big push

.Rbuildignore

@@ -11,3 +11,4 @@
 ^\.github$
 ^CITATION\.cff$
 ^README\.Rmd$
+^vignettes/articles$

.gitignore

@@ -4,3 +4,4 @@
 .Ruserdata
 docs
 inst/doc
+*.DS_Store

DESCRIPTION

@@ -17,15 +17,20 @@ LazyData: true
 Suggests: 
     knitr,
     rmarkdown,
-    testthat (>= 3.0.0)
+    testthat (>= 3.0.0),
+    here,
+    readxl,
+    stringr,
+    tinyplot,
+    tinytable,
+    ggplot2,
+    patchwork
 Config/testthat/edition: 3
 Imports: 
     cli,
     dplyr,
     rlang,
     tidyr,
-    tinyplot,
-    tinytable,
     vroom,
     zoo
 RoxygenNote: 7.3.2

NAMESPACE

@@ -1,13 +1,12 @@
 # Generated by roxygen2: do not edit by hand
 
-export(augment_trackball)
-export(mode)
-export(read_trackball_data)
+export(compute_kinematics)
+export(read_trackball)
+export(smooth_track)
 import(dplyr)
 importFrom(cli,cli_abort)
 importFrom(rlang,":=")
 importFrom(rlang,.data)
 importFrom(stats,median)
-importFrom(tidyr,replace_na)
 importFrom(vroom,vroom)
 importFrom(zoo,rollmean)

R/compute_kinematics.R

@@ -0,0 +1,44 @@
+#' Compute kinematics
+#'
+#' @description
+#' Compute kinematics.
+#'
+#' @param data Data frame
+#'
+#' @return An data frame data frame with kinematics computed
+#' @export
+#' @import dplyr
+#' @importFrom rlang :=
+#' @importFrom rlang .data
+#'
+compute_kinematics <- function(
+    data
+) {
+  # We first temporarily back-compute from our xy coordinates to the distances (dx, dy) covered between each observation (which is what we got from the sensors initially)
+  data <- data |>
+    dplyr::mutate(dx = .data$x - lag(.data$x),
+                  dy = .data$y - lag(.data$y),
+                  dt = .data$time - lag(.data$time))
+
+  # Find the sampling rate
+  sampling_rate <- round(1 / median(data$dt, na.rm = TRUE))
+
+  # Compute kinematics
+  data <- data |>
+    dplyr::mutate(distance = if_else(.data$dx^2 > 0 & .data$dy^2 > 0, sqrt(.data$dx^2 + .data$dy^2), 0),
+                  v_translation = .data$distance * sampling_rate,
+                  direction = if_else(.data$dx^2 > 0 & .data$dy^2 > 0, atan2(.data$dx, .data$dy), 0),
+                  direction = if_else(.data$dy == 0 | .data$dy == 0, NA, direction),
+                  rotation = direction - lag(direction),
+                  v_rotation = .data$rotation * sampling_rate,
+                  # We change the directions to stay within 2pi only here, otherwise rotation becomes harder to alculate
+                  direction = if_else(.data$direction < 0, .data$direction + 2*pi, .data$direction), # Keep direction between 0 and 2*pi
+                  direction = zoo::na.locf(.data$direction, na.rm = FALSE),
+    )
+
+  # Remove leftover columns
+  data <- data |>
+    select(-c("dx", "dy", "dt"))
+
+  return(data)
+}

R/compute_statistics.R

@@ -0,0 +1,42 @@
+#' Compute summary statistics
+#'
+#' @description
+#' Compute summary statistics for tracks
+#'
+#' @param data A kinematics data frame
+#' @param threshold_velocity Choose which observations to use based on the velocity. A number, "auto" or "none". Can take a number (e.g. estimated from histograms) or "auto". "auto" fits a density function to the velocities and tries to identify a local minimum between the first and second local maxima, and uses that as the threshold. "none" keeps all observations.
+#'
+#' @return An data frame data frame with kinematics computed
+#' @export
+#' @import dplyr
+#' @importFrom rlang :=
+#' @importFrom rlang .data
+#'
+compute_statistics <- function(
+    data,
+    threshold_velocity
+) {
+  # Make some tests to ensure that `compute_kinematics` has been run first
+
+  # Make sure to remove observations with almost no movement (figure out a robust method for this)
+
+  # Compute translational and rotational separately (maybe?) and gather at the end
+  data <- data |>
+    dplyr::summarise()
+
+    # dplyr::mutate(distance = sqrt(.data$dx^2 + .data$dy^2),
+    #               v_translation = .data$distance * sampling_rate,
+    #               direction = atan2(.data$dx, .data$dy),
+    #               direction = if_else(.data$dy == 0 | .data$dy == 0, NA, direction),
+    #               direction = if_else(.data$direction < 0, .data$direction + 2*pi, .data$direction), # Keep direction between 0 and 2*pi
+    #               direction = zoo::na.locf(.data$direction, na.rm = FALSE),
+    #               rotation = direction - lag(direction),
+    #               # rotation = dplyr::if_else(abs(.data$dx) > 1 & abs(.data$dy) > 1,
+    #               #                           abs(dplyr::lag(.data$direction) - .data$direction),
+    #               #                           0),
+    #               # rotation = dplyr::if_else(.data$rotation > pi, 2*pi - .data$rotation, .data$rotation),
+    #               v_rotation = .data$rotation * sampling_rate,
+    # )
+
+  return(data)
+}