如何计算r中分组数据帧中每列中的观察次数

Question

I have a dataframe composed of geochemical samples result which includes the following variables:

Year, Zone, *48 analyzed elements*, *more information*.

I want to know how many samples were collected each year, in each zone for every elements. So basically, I would like a table that would look like this:

Year,Zone,Ag_ppm, ..., Zr_ppm
1981, ZoneA, 0, ..., 0 
1981, ZoneB, 20, ..., 0
1983, ZoneA, 0, ..., 150 

I have tried the following:

 Elt_count <- SoilGeology %>%
  group_by(Year, Zone) %>%
  summarise_at(vars(Ag_ppm:Zr_ppm),funs(sum)) %>%
  select(Year, Zone, Ag_ppm:Zr_ppm)

It works but it does not give me the information I want (I don't want the cummulative sum of the samples , but a count of every samples). I have then tried:

Elt_count <- SoilGeology %>%
  group_by(Year, Zone) %>%
  summarise_at(vars(Ag_ppm:Zr_ppm),funs(n)) %>%
  select(Year, Zone, Ag_ppm:Zr_ppm)

But I get the following error: Error in summarise_impl(.data, dots) : n() does not take arguments

I have also tried:

d <- SoilGeology %>%
  group_by(Year, Zone) %>%
  summarise_all(n) %>%
  select(Year, Zone, Ag_ppm:Zr_ppm)

But I get the same error as above: Error in summarise_impl(.data, dots) : n() does not take arguments

And also tried with count:

Elt_count <- SoilGeology %>%
  group_by(Year, Zone) %>%
  count(Au_ppm:Zr_ppm, na.rm = TRUE) %>%
  select(Year, Zone, Ag_ppm:Zr_ppm)

But, I get the error:

Error in mutate_impl(.data, dots) : Evaluation error: NA/NaN argument.
In addition: Warning messages:
1: In Au_ppm:Zr_ppm :
  numerical expression has 52 elements: only the first used
2: In Au_ppm:Zr_ppm :
  numerical expression has 52 elements: only the first used

Does someone has an explanation for these errors? Or a better solution for my problem?

Thanks!

Answer 1

Perhaps the following code is what you desire.

library(dplyr) count(SoilGeology, year, zone, Ag_ppm:Zr_ppm)

Or equivalently with the pipe function

SoilGeology %>% count(SoilGeology, year, zone, Ag_ppm:Zr_ppm)

Alternately,

SoilGeology %>% group_by(year, zone, Ag_ppm:Zr_ppm) %>% summarise(number = n())

Or

SoilGeology %>% group_by(year, zone, Ag_ppm:Zr_ppm) %>% tally()

If errors persist, the 'class()' of your variables should be checked. Values may need to be coerced to numeric. If needed, try variable -> as.numeric(variable) and try again.

Answer 2

sum adds numbers, n() and count() count rows. If numbers greater than 0 have special meaning for you, you need to tell R that. The classic way to count the number of things meeting a condition is sum(..test for condition..) , so if you want the number of elements of x that are greater than 0 , sum(x > 0) will do it. This is the function you want to apply to all columns:

# reproducible example on built-in data
mtcars %>%
  group_by(cyl) %>%
  summarize_at(vars(disp:carb), function(x) sum(x > 5))

# for your data
Elt_count <- SoilGeology %>%
  group_by(Year, Zone) %>%
  summarise_at(vars(Ag_ppm:Zr_ppm), function(x) sum(x > 0))

I don't know your data. You may want to change it to sum(x != 0) if there are negative numbers you want to count too. If there are missing values, sum(x > 0, na.rm = TRUE) (if you look at ?sum , it does take a na.rm argument).