Context : I have some spatial point data (ie lon/lat coordinates), and each point is associated with a date. I've clustered points that are close together, but I now want to split these clusters into groups so that if sorted by date the clusters are sequential and grouped together. Dates can have gaps, and I only want to slit when an observation fully divides a group, ie it's not just on the edge
Essentially, given the below cluster
and day
fields I want to generate desired
.
cluster day desired
1 1 1 1
2 1 1 1
3 1 2 1
4 1 4 1
5 2 6 2
6 2 7 2
7 2 8 2
8 1 8 3
9 3 9 4
10 3 12 4
11 3 12 4
12 2 12 5
13 2 14 5
14 3 18 6
15 3 19 6
Here's a complete example, note that the spatial coordinates are essentially irrelevant, I've just included them for completeness. Also, in my actual dataset day
is a date object, but I've used an integer for simplicity.
library(ggplot2)
pts <- data.frame(rbind(
cbind(lon = rnorm(5, 0, 0.1), lat = rnorm(5, 0, 0.1),
day = c(1, 1, 2, 4, 8)),
cbind(lon = rnorm(5, 1, 0.1), lat = rnorm(5, 1, 0.1),
day = c(6, 7, 8, 12, 14)),
cbind(lon = rnorm(5, 1, 0.1), lat = rnorm(5, 0, 0.1),
day = c(9, 12, 12, 18, 19))
))
hc <- hclust(dist(pts[c("lon", "lat")]))
pts$cluster <- cutree(hc, k = 3)
ggplot(pts) +
geom_text(aes(lat, lon, label = day, col = as.factor(cluster)))
The grouping I want is this:
pts$desired <- c(1, 1, 1, 1, 3,
2, 2, 2, 5, 5,
4, 4, 4, 6, 6)
ggplot(pts) +
geom_text(aes(lat, lon, label = day, col = as.factor(desired)))
This solution comes courtesy of @docendodiscimus in the comments to the original question.
library(dplyr)
pts <- pts %>%
arrange(day, desc(cluster)) %>%
mutate(new_cluster = cumsum(c(1L, diff(cluster) != 0)))
all.equal(pts$desired, pts$new_cluster)
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.