R finding the first value in a data frame that falls within a given threshold

Question

I am a fairly new user and I need your help with a task that I am stuck on. If my question has been asked/answered before I would be grateful if you could kindly guide me to the relevant page.

I have the following data set (lbnp_br) which is optical density (OD) measured over time (in seconds):

 time   OD
1891    -244.6
1891.5  -244.4
1892    -242
1892.5  -242
1893    -241.1
1893.5  -242.4
1894    -245.2
1894.5  -249.6
**1895  -253.9**
1895.5  -254.5
1896    -251.9
1896.5  -246.7
1897    -242.4
1897.5  -234.6
1898    -225.5

I need to find out how responsive the study device is by measuring how long it takes to reach the threshold for optical density.

For this I have calculated the coefficient of variation (CV) of OD and I am using mean OD (-252.9098) +/- 2*CV to define a response threshold. For the above data the threshold is set as (mean OD + 2*CV = -252.9917), and (mean OD - 2*CV = -252.8278).

I now need to calculate the time in seconds from the start (1891 seconds) to the first OD value that exceed the +/- threshold values. For example for the above data frame this threshold is exceeded at 1895 seconds corresponding to an OD of -253.9.

I now have to repeat this 3 times for each study subject and 17 subjects overall, thus, I am looking for a function where I can define the data frame and the threshold values, and it will return the first OD value where it exceeds the defined thresholds (all_threshold$sup_2_minus) and (all_threshold$sup_2_plus) and its corresponding time.

I have tried subset a advised elsewhere:

subset(lbnp_br, lbnp_br$OD < all_threshold$sup_2_minus & lbnp_br$OD > all_threshold$sup_2_plus)  

However, this doesn't return what I am looking for.

and also

ifelse(lbnp_br$OD > all_threshold$sup_2_plus & lbnp_br$OD < all_threshold$sup_2_minus, lbnp_br$OD, NA)

which returns NA and doesn't specify the exact value of OD and the time.

Answer 1

This is not a short answer, but hopefully clear. It uses the dplyr package:

library(dplyr)

find_time = function(df, threshold){
  return_value = df %>%
    arrange(time) %>%
    filter(OD < threshold) %>%
    slice(1)
  return(return_value)
}

find_time(data, threshold)

This will sort (arrange) your data based on time, subset (filter) your data for values of OD below the threshold, take the first value (slice), and return it.

Answer 2

A one liner:

function (dfr, threshold) dfr$OD[ min(which(dfr$OD > threshold)) ]

Gives a warning and NA if there is no such row in the data frame, which is probably what you want.

An alternative, purrr -based solution:

function (dfr, threshold) purrr::detect(dfr$OD, ~ .x > threshold)

which returns NULL if nothing is found, more correct I guess.

Answer 3

Using the above code, I added a few extra conditions to get exactly what I was looking for and here it is for anyone who may need something similar:

find_time <- function(df, df2, df3, threshold_1, threshold_2, threshold_3, threshold_4, threshold_5, threshold_6){
  return_value_1 = df %>%
    arrange(time) %>%
    filter(OD > threshold_1) %>%
    slice_(1)
  colnames(return_value_1)[1] <- "time_hdt_upper"
  colnames(return_value_1)[2] <- "OD_hdt_upper"

  if (nrow(return_value_1) == 0) {
    return_value_1[1,1] <- NA
    return_value_1[1,2] <- NA
  }


  return_value_2 = df %>%
    arrange(time) %>%
    filter(OD < threshold_2) %>%
    slice_(1)
  colnames(return_value_2)[1] <- "time_hdt_lower"
  colnames(return_value_2)[2] <- "OD_hdt_lower"

  if (nrow(return_value_2) == 0) {
    return_value_2[1,1] <- NA
    return_value_2[1,2] <- NA
  }

  return_value_3 = df2 %>%
    arrange(time) %>%
    filter(OD > threshold_3) %>%
    slice_(1)
  colnames(return_value_3)[1] <- "time_lbnp_upper"
  colnames(return_value_3)[2] <- "OD_lbnp_upper"

  if (nrow(return_value_3) == 0) {
    return_value_3[1,1] <- NA
    return_value_3[1,2] <- NA
  }


  return_value_4 = df2 %>%
    arrange(time) %>%
    filter(OD < threshold_4) %>%
    slice_(1)
  colnames(return_value_4)[1] <- "time_lbnp_lower"
  colnames(return_value_4)[2] <- "OD_lbnp_lower"

  if (nrow(return_value_4) == 0) {
    return_value_4[1,1] <- NA
    return_value_4[1,2] <- NA
  }



  return_value_5 = df3 %>%
    arrange(time) %>%
    filter(OD > threshold_5) %>%
    slice_(1)
  colnames(return_value_5)[1] <- "time_hut_upper"
  colnames(return_value_5)[2] <- "OD_hut_upper"

  if (nrow(return_value_5) == 0) {
    return_value_5[1,1] <- NA
    return_value_5[1,2] <- NA
  }



  return_value_6 = df3 %>%
    arrange(time) %>%
    filter(OD < threshold_6) %>%
    slice_(1)
  colnames(return_value_6)[1] <- "time_hut_lower"
  colnames(return_value_6)[2] <- "OD_hut_lower"

  if (nrow(return_value_6) == 0) {
    return_value_6[1,1] <- NA
    return_value_6[1,2] <- NA
  }



  return(data.frame(return_value_1, return_value_2, return_value_3, return_value_4, return_value_5, return_value_6))


}

which gives

find_time_threshold <- find_time(hdt_br, lbnp_br, hut_br, all_threshold$base_plus, all_threshold$base_minus, all_threshold$sup_2_plus, all_threshold$sup_2_minus, all_threshold$sup_3_plus, all_threshold$sup_3_minus)
> find_time_threshold

  time_hdt_upper OD_hdt_upper time_hdt_lower OD_hdt_lower time_lbnp_upper OD_lbnp_upper time_lbnp_lower
1          596.5        123.3            506         91.3              NA            NA            1706
  OD_lbnp_lower time_hut_upper OD_hut_upper time_hut_lower OD_hut_lower
1        -27.89         3186.5       -82.98           2909       -211.7