r - R循环黄土功能超过24小时（一天）

Question

我有几个小时和几天的免费停车位数据。

这是100个随机样本。

sl <- list(EmptySlots = c(7, 6, 20, 5, 16, 20, 24, 5, 24, 24, 15, 11, 
8, 6, 13, 2, 21, 6, 1, 6, 9, 1, 8, 0, 20, 9, 20, 11, 22, 24, 
1, 2, 12, 6, 8, 2, 23, 18, 8, 3, 20, 2, 1, 0, 5, 21, 1, 4, 20, 
15, 24, 12, 4, 14, 2, 4, 20, 16, 2, 10, 2, 1, 24, 9, 22, 7, 6, 
3, 20, 13, 1, 16, 12, 5, 2, 7, 4, 1, 6, 1, 1, 2, 0, 13, 24, 6, 
13, 7, 24, 24, 15, 6, 10, 1, 2, 9, 5, 2, 11, 15), hour = c(8, 
16, 23, 14, 18, 7, 17, 15, 19, 19, 17, 17, 16, 14, 17, 12, 19, 
10, 10, 13, 16, 10, 16, 11, 12, 9, 0, 15, 16, 21, 10, 11, 17, 
11, 16, 15, 23, 7, 16, 14, 18, 14, 14, 9, 15, 2, 10, 9, 19, 17, 
20, 16, 12, 17, 12, 9, 23, 9, 15, 17, 10, 12, 18, 17, 18, 17, 
13, 10, 7, 8, 10, 18, 11, 11, 12, 17, 12, 9, 14, 15, 10, 11, 
10, 10, 20, 16, 18, 15, 21, 18, 17, 13, 8, 11, 15, 16, 11, 9, 
12, 18))

一种通过 快速计算 LOESS 函数的方法ggplot2。

sl <- as.data.frame(sl)
library(ggplot2)
qplot(hour, EmptySlots, data=sl, geom="jitter") + theme_bw() + stat_smooth(size = 2)

告诉 LOESS 函数 0 和 24 是邻居的最佳方法是什么？即，如果我们以这种方式估计，左边和右边的线应该是相同的值。

关于从哪里开始的指针会很好。

Answer 1

我很想在两边复制数据：

library(ggplot2)
empty <- c(7, 6, 20, 5, 16, 20, 24, 5, 24, 24, 15, 11, 8, 6, 13, 2, 21, 6, 1, 6, 9, 1, 8, 0, 20, 9, 20, 11, 22, 24, 1, 2, 12, 6, 8, 2, 23, 18, 8, 3, 20, 2, 1, 0, 5, 21, 1, 4, 20, 15, 24, 12, 4, 14, 2, 4, 20, 16, 2, 10, 2, 1, 24, 9, 22, 7, 6, 3, 20, 13, 1, 16, 12, 5, 2, 7, 4, 1, 6, 1, 1, 2, 0, 13, 24, 6, 13, 7, 24, 24, 15, 6, 10, 1, 2, 9, 5, 2, 11, 15)
hour <- c(8, 16, 23, 14, 18, 7, 17, 15, 19, 19, 17, 17, 16, 14, 17, 12, 19, 10, 10, 13, 16, 10, 16, 11, 12, 9, 0, 15, 16, 21, 10, 11, 17, 11, 16, 15, 23, 7, 16, 14, 18, 14, 14, 9, 15, 2, 10, 9, 19, 17, 20, 16, 12, 17, 12, 9, 23, 9, 15, 17, 10, 12, 18, 17, 18, 17, 13, 10, 7, 8, 10, 18, 11, 11, 12, 17, 12, 9, 14, 15, 10, 11, 10, 10, 20, 16, 18, 15, 21, 18, 17, 13, 8, 11, 15, 16, 11, 9, 12, 18)
emptyrep <- rep.int(empty,3)
hourrep <- c(hour,hour+24,hour-24)
sl <- data.frame(empty=emptyrep, hour=hourrep)
qplot(hour, empty, data=sl, geom="jitter") + theme_bw() + geom_smooth(method="loess",size = 1.5,span=0.2) +  coord_cartesian(xlim=c(0,24))

...就像 joran 几分钟前所说的（呜呜）