R 图密度平滑时间序列

发布于 2024-10-16 17:05:05 字数 422 浏览 2 评论 0原文

我希望对一些时间序列数据进行概率分布。我的数据采用以下格式

00:00, 3
01:00, 50
05:00, 13
10:00, 34
17:00, 80
21:00, 100

时间列有一些缺失值，R 必须对其进行插值。我想要一条漂亮的平滑曲线来突出繁忙的时期。我尝试过 ts、密度和绘图，但这些并没有产生我想要的结果。例如，

data1 <- read.csv(file="c:\\abc\\ts.csv", head=FALSE, sep=",")
data1$V1 <- strptime(data1$V1, format="%H:%M")
plot(data1$V2, density(data1$V1), type="l")

但这给了我以疯狂的顺序和概率分布绘制的线条。

原文

I wish to make a probability distribution of some time series data. My data is in the following format

00:00, 3
01:00, 50
05:00, 13
10:00, 34
17:00, 80
21:00, 100

The time column has some missing values that R will have to interpolate. I want to get a nice smooth curve to highlight the busy periods. I have tried with ts, density and plot but these don't produce what I'm after. For example,

data1 <- read.csv(file="c:\\abc\\ts.csv", head=FALSE, sep=",")
data1$V1 <- strptime(data1$V1, format="%H:%M")
plot(data1$V2, density(data1$V1), type="l")

But this gives me lines drawn in crazy order and as a probability distribution.

分享到QQ

分享到微博

如果你对这篇内容有疑问，欢迎到本站社区发帖提问参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。

发布评论

需要登录才能够评论，你可以免费注册一个本站的账号。

琴流音 2024-10-23 17:05:05

我想你肯定在寻找包zoo，它有几个功能来处理 NA。另请参阅 na.aggregate、na.approx 和 na.locf。

回复收藏 0 原文

溺深海 2024-10-23 17:05:05

你让事情变得比你想象的要困难一些。我现在会在您的时间前面添加一个日期，让事情变得更容易。

另外，我添加了一个变量“texinp”和一个 textConnection() 语句，以便您可以剪切/粘贴以下代码并直接运行它。数据被加载到变量 texinp 中，并由 read.zoo 语句以与读取 .csv 文件类似的方式读取。现在，这将允许您绘制内容并让您了解如何使用 read.zoo 读取 .csv 文件。

library(zoo)
library(chron)

texinp <- "
Time,  Mydata
2011-02-06 00:00, 3
2011-02-06 01:00, 50
2011-02-06 05:00, 13
2011-02-06 10:00, 34
2011-02-06 17:00, 80
2011-02-06 21:00, 100"

myd.zoo <- read.zoo(textConnection(texinp), header=TRUE, FUN = as.chron, sep=",")
myd.zoo

plot(myd.zoo)

从你的问题来看，你谈到了“繁忙期”。我可能是错的，但我假设 21:00 时的值 100 是“最繁忙的时段”。如果这是真的，那么您不需要密度图，上面的图就是您想要的。

如果我错了请告诉我。

You made it a little harder than you might realize. I'll make it easier for now by adding a date in front of your times.

Also, I added a variable "texinp" and a textConnection() statement so you can cut/paste the following code and run it directly. The data is loaded into variable texinp and is read by the read.zoo statement in a similar way to reading a .csv file. For now, this will allow you to plot things and gives you an idea of how to read .csv files using read.zoo.

library(zoo)
library(chron)

texinp <- "
Time,  Mydata
2011-02-06 00:00, 3
2011-02-06 01:00, 50
2011-02-06 05:00, 13
2011-02-06 10:00, 34
2011-02-06 17:00, 80
2011-02-06 21:00, 100"

myd.zoo <- read.zoo(textConnection(texinp), header=TRUE, FUN = as.chron, sep=",")
myd.zoo

plot(myd.zoo)

From your question, you talked about "busy periods". I may be wrong, but I'm assuming that the value of 100 at time 21:00 is the "busiest period". If that's true, then you don't need a density plot, and the above plot is what you're after.

Let me know if I'm wrong.

回复收藏 0 原文

~没有更多了~