Plotting a Mean Heart Rate Over Specific Time Points

[nviet]

I'm expanding on a previous conversation (Reference to other post). I'm having trouble figuring out how to grab a mean for a group of specific ID numbers across different units of time within my data. For example, if I wanted to take CD-01 through CD-09 & CD-11 across V2 to V6 and plot that mean, then take V7 through V10 and plot that then V11 and V12 and the same thing. Then do that again for CD-103 and 105 and the SS etc. I suppose I still have to keep the data frame in long but I'm struggling with the data management part like selecting the columns.

So basically a mean plot that across at different points in the data set. Any information would be helpful. Thank you

heartrate <- tibble::tribble(
                       ~V1,  ~V2,  ~V3,  ~V4,  ~V5,  ~V6,  ~V7,  ~V8,  ~V9, ~V10, ~V11, ~V12,
               "ID Number",   0L,   1L,   2L,   3L,   4L,   5L,   6L,   7L,   8L,   9L,  10L,
                   "CD-01",  66L,  66L,  64L,  64L,  58L,  58L,  58L,  57L,  56L,  56L,  57L,
                   "SS-02",  85L,  84L,  83L,  81L,  80L,  79L,  78L,  78L,  79L,  80L,  80L,
                   "CD-03", 103L, 103L, 103L, 103L, 103L, 103L, 103L, 103L, 103L, 103L, 103L,
                   "CD-04", 115L, 115L, 114L, 114L, 114L, 114L, 112L, 111L, 110L, 109L, 109L,
                  "SS-101",  63L,  63L,  63L,  63L,  77L,  78L,  80L,  77L,  76L,  67L,  67L,
                   "CD-05",  84L,  63L,  62L,  62L,  63L,  58L,  58L,  54L,  62L,  66L,  66L,
                  "SS-102",  49L,  48L,  46L,  46L,  45L,  46L,  45L,  45L,  45L,  45L,  46L,
                  "CD-103",  70L,  69L,  68L,  68L,  69L,  69L,  69L,  69L,  68L,  69L,  70L,
                   "CD-06",  83L,  84L,  84L,  84L,  84L,  84L,  82L,  82L,  82L,  82L,  81L,
                   "SS-08",  79L,  78L,  67L,  67L,  68L,  62L,  62L,  62L,  63L,  72L,  72L,
                   "CD-09", 240L, 202L, 141L, 106L,  89L,  74L,  71L,  71L,  71L,  72L,  74L,
                   "SS-10",  78L,  78L,  77L,  76L,  75L,  75L,  75L,  75L,  76L,  76L,  76L,
                  "CD-105",  66L,  66L,  66L,  66L,  66L,  66L,  67L,  67L,  67L,  67L,  68L,
                   "CD-11",  66L,  66L,  76L,  76L,  76L,  77L,  75L,  75L,  75L,  75L,  76L
               )
head(heartrate)

[xvalda]

Hi @nviet
I'll answer at least the first part of your question.
There are a few concepts of tidy data to address:

• observations (time element) must be in rows, not columns
• variables (your categories) must be in columns, not in rows
  --> so we need to transpose
• since categories ((CD-01, SS-02, ...) will become variables, they need to be valid R variable names to avoid using backticks

#create a named vector to clean up variable names, in a standard tidyverse format
# this will be used in the second code chunk
var_names <- heartrate %>% transmute(clean_names = str_replace(V1, "[ -]", "_"), old_names = c(1:nrow(.))) %>% 
  deframe()
# transform your initial tibble into a tidy format
heartrate2 <- heartrate %>% 
  # transpose (1 observation per row, one column per category) and 
  t() %>%
  # transposing turns your tibble into a matrix, it needs to be re-tibbled
  as_tibble() %>% 
  # changing variable names from the name vector
  # more info here if you'd like: https://forum.posit.co/t/rename-with-a-named-vector-list-and-contribution-to-the-tidyverse/2383
  rename(!!! var_names) %>% 
  # remove first row that was "pushed" to variable names
  slice(-1) %>% 
  # transposing your initial tibble that contained all integers except the V1 variable coerced the matrix to be characters
  # now that we have no character variable in the tibble, we convert all to integer again
  mutate_all(as.integer) %>% 
  # I add a new "time" column that will help when plotting (personal choice)
  mutate(time = str_c("t", 1:nrow(.))) %>% 
  # convert to long format that is ggplot-friendly
  pivot_longer(CD_01:CD_11, names_to = "category", values_to = "value") 

# now you can plot the usual way
heartrate2 %>% 
  ggplot(aes(x = fct_reorder(time, ID_Number), y = value, group = category, color = category)) + 
  geom_line() + 
  labs(x = "time", y = "heart rate")

# or alternatively
heartrate2 %>% 
  ggplot(aes(x = fct_reorder(time, ID_Number), y = value, group = category, color = category)) + 
  geom_line() + 
  labs(x = "time", y = "heart rate") + 
  facet_wrap(~category)

I hope this helps with part of your question.

[nviet]

Thank you for your help but this doesn't do much to address the question. If you look at the reference to the other post this was kind of addressed already. I sincerely appreciate it though.

[xvalda]

Last attempt before the weekend, I hope this is closer to what you're loking for, grouped the categories into 4 main groups, calculate mean of each per time time element and plotted:

heartrate3 <- heartrate %>% 
  t() %>%
  as_tibble() %>% 
  rename(!!! var_names) %>% 
  slice(-1) %>% 
  mutate_all(as.integer) %>% 
  mutate(time = str_c("t", 1:nrow(.)), .before = ID_Number) %>%
  pivot_longer(CD_01:CD_11, names_to = "category", values_to = "value") %>% 
  mutate(main_group = case_when(
    str_detect(category, "CD_0") ~ "CD_0x",
    str_detect(category, "CD_1") ~ "CD_1x",
    str_detect(category, "SS_0") ~ "SS_0x",
    str_detect(category, "SS_1") ~ "SS_1x"
  )) %>% 
  group_by(main_group, time) %>% 
  summarize(mean_group = mean(value)) %>% 
  mutate(ID_Number = as.numeric(str_remove(time, "t")))

heartrate3 %>% 
  ggplot(aes(x = fct_reorder(time, ID_Number), y = mean_group, group = main_group, color = main_group)) + 
  geom_line() + 
  labs(x = "time", y = "heart rate")

I hope this is getting closer.

[nviet]

This might be it.

The "str_detect(category, "CD_0")~"CD_0x", I assuming (in layman terms) selected all the cases with CD_0 to CD_0X(X being the maximum number in the data set as long as there is a "CD_0" in front of the following number)?

[xvalda]

Yes, that's the spirit, for example str_detect("CD_0656468154dofjhosf", "CD_0") will return TRUE, thus validating the first condition in the case_when statements.
I added something in the case_when function, when a category doesn't match any of the 4 patterns we defined, then it will have a fallback value of "other category" that will appear in your plot.
I copied the whole block below with the add-on.

Also note that what is in the str_detect() arguments is a very basic regex that works on the data that you shared, if you have much more data with different patterns, it could be worth making the expressions a bit more solid. But this part you will be able to diagnose if you ever see the "other category" line in your plot.

heartrate3 <- heartrate %>% 
  t() %>%
  as_tibble() %>% 
  rename(!!! var_names) %>% 
  slice(-1) %>% 
  mutate_all(as.integer) %>% 
  mutate(time = str_c("t", 1:nrow(.)), .before = ID_Number) %>%
  pivot_longer(CD_01:CD_11, names_to = "category", values_to = "value") %>% 
  mutate(main_group = case_when(
    str_detect(category, "CD_0") ~ "CD_0x",
    str_detect(category, "CD_1") ~ "CD_1x",
    str_detect(category, "SS_0") ~ "SS_0x",
    str_detect(category, "SS_1") ~ "SS_1x", 
    TRUE ~ "other category"
  )) %>% 
  group_by(main_group, time) %>% 
  summarize(mean_group = mean(value)) %>% 
  mutate(ID_Number = as.numeric(str_remove(time, "t")))