I have data relating to 36 regions of interest (ROI), approx. 380 pixels per ROI. My data is like:
ROI_name T_K
1 bt_full_05 303.88
1.1 bt_full_05 303.93
1.2 bt_full_05 303.72
1.3 bt_full_05 303.43
1.4 bt_full_05 302.93
1.5 bt_full_05 302.93
...
36.362 bt_full_40 301.65
36.363 bt_full_40 301.47
36.364 bt_full_40 301.52
36.365 bt_full_40 302.02
36.366 bt_full_40 303.28
36.367 bt_full_40 303.78
I want to compute mean T_K for each ROI, but filter out values below a given quantile, eg 0.25, and output the mean of values up to that quantile. Ideally I could report mean T_K for several quantiles, 0.1, 0.25, 0.5... I have:
groupquant <- cleared_data %>% group_by(ROI_name) %>%
summarise(quants = quantile(T_K, 0.1))
which gives me the quantiles. But this
groupquant <- cleared_data %>% group_by(ROI_name) %>%
filter(cleared_data$T_K <= quantile(T_K, 0.1)) #%>%
throws
Error: Result must have length 392, not 14082
I'm getting nowhere! Cheers, Andrew.
I think the sample data is a bit small to demonstrate that you wanna do. So I created my own data, which is called foo
. For each ROI_name
, I removed some data with filter()
. All values which are smaller than quantile(T_K, 0.25)
are removed. Then, I decided to take values for two quantile points (ie, 0.5 and 0.75). In the summarize()
part, I am getting a vector with two numeric values and creating a data frame for each group. Finally, I used unnest()
to create the final output.
library(tidyverse)
set.seed(111)
foo <- tibble(ROI_name = rep(c("bt_full_05", "bt_full_40", "bt_full_2"), each = 30),
T_K = runif(n = 90, min = 0, max = 300))
group_by(foo, ROI_name) %>%
filter(T_K > quantile(T_K, 0.25)) %>%
summarize(temp = list(enframe(quantile(x = T_K, prob = c(0.5, 0.75)),
name = "percentile"))) %>%
unnest(temp)
ROI_name percentile value
<chr> <chr> <dbl>
1 bt_full_05 50% 157.
2 bt_full_05 75% 183.
3 bt_full_2 50% 157.
4 bt_full_2 75% 229.
5 bt_full_40 50% 192.
6 bt_full_40 75% 237.
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.