数据框具有3列,即id,days和sum。我想生成一个总和的热图,y轴为id,x轴为天。问题在于数据稀疏,因此热图由离散条组成。我希望条形图向右延伸,以使条形图变为实线,并在总和更改值时更改颜色,并保持该颜色直到第二天的值右侧。
这是一个生成我正在制作的情节类型的示例。
library(ggplot2)
set.seed(13)
x_id <- sample( LETTERS[1:5], 100, replace=TRUE,
prob=c(0.15, 0.2, 0.35, 0.1, 0.2) )
x_sum <- sample( c(5, 30, 60, 120, 180, 240, 360), 100, replace=TRUE,
prob=c(.1, .1, .2, .2, .2, .1, .1) )
x_days <- sample.int(2000, 100, replace = TRUE)-1000
df <- data.frame(id = x_id, Days = x_days, sum = x_sum)
ggp <- ggplot(data = df,
mapping = aes(x = Days,
y = id,
fill = sum)) +
geom_tile() +
xlab(label = "Days") + ylab(label = 'id') +
scale_fill_gradient(low = "blue", high = "red")
print(ggp)
我希望颜色向右延伸。我认为这意味着数据框应按id和days排序,并且必须为每个id添加其他行,以便用sum和id的值等于sum / id的最后一个值来填写缺少的日期。但是,如何为每个ID添加行并填写缺少的值?最右边的颜色应延长固定的长度,这样颜色才更可见,例如延长30天。
此外,颜色图显示指示临界值。假设临界值为180。然后,对于从零到临界值(180)的总和,颜色应从绿色(0)变为黄色(179),对于大于临界值(180)的值,颜色应为浅红色(180)到深红色(最大值或360)
更新:
这是一种填充稀疏矩阵的解决方案
library(tidyr)
setkey(DT, id, Days)
DT_fill_NA <- DT[setkey(DT[, .(min(Days):(max(Days)+30)), by = id], id, V1)]
DT_fill <- fill(DT_fill_NA, c('sum'), .direction = "down")
ggp <- ggplot(data = DT_fill,
mapping = aes(x = Days,
y = id,
fill = sum)) +
geom_tile() +
xlab(label = "Days") + ylab(label = 'id') +
scale_fill_gradient(low = "blue", high = "red")
print(ggp)
这将创建具有稀疏条的图形,该稀疏条向右延伸到下一个条
现在应修改颜色图以指示临界值。设临界值为180。然后,对于从零到临界值(180)的和,颜色应从绿色(0)变为黄色(179),对于大于临界值(180)的和,颜色应从浅红色(180)到深红色(最大值或360)
第二次更新
一种在180处中断时生成绿色的方法如下
ggp <- ggplot(data = DT_fill,
mapping = aes(x = Days,
y = id,
fill = sum)) +
geom_tile() +
xlab(label = "Days") + ylab(label = 'id') +
scale_fill_gradient2(low = "green", mid = "indianred2", high = "red2",
midpoint = 180, breaks = c(50, 100, 200, 300)) +
theme_bw()
print(ggp)
我不确定这是否清楚地将断点标识为特定值。怎样才能使绿色/红色之间的临界点正确地设置为临界值(180)?
答案 0 :(得分:0)
这是一种从稀疏矩阵中突出显示临界值的填充热图的方法。
library(ggplot2)
library(data.table)
library(tidyr)
set.seed(13)
n_rows = 200
x_id <- sample( LETTERS[1:5], n_rows, replace=TRUE,
prob=c(0.15, 0.2, 0.35, 0.1, 0.2) )
x_sum <- sample( c(0, 5, 30, 60, 120, 180, 240, 270, 360), n_rows, replace=TRUE,
prob=c(.05, .05, .1, .2, .2, .2, .1, 05, .05) )
x_days <- sample.int(2000, n_rows, replace = TRUE)-1000
DT <- data.table(id = x_id, Days = x_days, sum = x_sum)
setkey(DT, id, Days)
DT_fill_NA <- DT[setkey(DT[, .(min(Days):(max(Days)+100)), by = id], id, V1)]
DT_fill <- fill(DT_fill_NA, c('sum'), .direction = "down")
brks = c(-1, 50, 100, 180, 250, 300, max(DT_fill$sum))
DT_fill$sum_factors = cut(DT_fill$sum, breaks = brks, ordered_result = TRUE, right = TRUE)
unique(DT_fill$sum_factors)
ggp <- ggplot(data = DT_fill,
mapping = aes(x = Days,
y = id,
fill = sum_factors)) +
geom_tile() +
xlab(label = "Days") + ylab(label = 'id') +
scale_fill_manual(values = c("green4", "green3", "green",
"firebrick1", "firebrick3", "firebrick4")) +
theme_bw()
print(ggp)