Data Science with R for the Life Sciences

Combining Concepts

Notes

  • Combining concepts is tricky and comes down to problem solving
  • Remember the rules for problem decomposition
    • Break thinks into a few parts
    • Keep breaking them down until each part seems doable
    • Code and test one part at time

Patterns

  • There are some common patterns than can be useful

Subsetting

library(dplyr)

data = data.frame(time = 1:100, runif(100, 1, 10))
window = 10
for (tmin in 1:max(data$time)-window) {
  window_data = filter(data, time >= tmin, time < tmin + window)
}

Nesting

data = data.frame(time = 1:100, runif(100, 1, 10))
windows = c(1, 5, 10, 20)
for (window in windows){
  for (tmin in 1:max(data$time)-window) {
    window_data = filter(data, time >= tmin, time < tmin + window)
  }
}

Counting

library(dplyr)

data = data.frame(time = 1:100, runif(100, 1, 10))
window = 10
windows_with_no_data = 0
for (tmin in 1:max(data$time)-window) {
  window_data = filter(data, time >= tmin, time < tmin + window)
  if (nrow(window_data) == 0){
      windows_with_no_data = windows_with_no_data + 1
  }
}

Nested conditionals

est_mass <- function(volume, veg_type, age){
  if (veg_type == "tree") {
    if (age < 5) {
	  mass <- 1.6 * volume^0.8
	  } else {
	  mass <- 2.65 * volume^0.9
	}
  } else if (veg_type == "grass" | veg_type == "shrub") {
	mass <- 0.65 * volume^1.2
  } else {
	print("I don't know how to convert volume to mass for that vegetation type")
	mass <- NA
  }
  return(mass)
}

est_mass(1.6, "tree", age = 2)
est_mass(1.6, "shrub", age = 5)
  • First checks if the vegetation type is “tree”
  • If it is checks to see if it is < 5 years old
  • If so does one calculation, if not does another
  • But nesting can be difficult to follow so try to minimize it