如何在 R 中将数据帧上传到 ndtv？

我的目标是使用 R 中的三个包进行动态可视化：ndtv, network, and networkDynamic包。

我创建了一个dataset根据此示例数据集排序的信息网络动态时间可视化研讨会（第 7 页。）

根据网络动态手册第49页，上传数据集并将其转换为networkDynamic对象的一种方法如下：

rawEdges<-read.table(paste(path.package("networkDynamic"),"/enron_timebased3.tsv", sep=''),header=TRUE)

但是，当我尝试运行时animation.render(rawEdges)

R 抛出错误消息：

第一个参数必须是网络对象。

为了解决这个问题，我创建了一个网络对象：net<-network(rawEdges)

and try:

animation.render(net, rawEdges)

新的错误消息：

'$tmp', "initial.coords", value = c(0, 0, 0, : 替换有 34 行，数据有 26)

有谁知道如何解决这一问题？

认为您的示例存在几个问题：

• 你需要创建一个networkDynamic对象，而不是网络 目的
• 您将必须进行一些时间格式转换 表正确解析，以及创建数字 ID
• 命令是render.animation() not animation.render()

首先，让我们设置一些可以加载的示例数据。只需要示例数据的前 4 列：

# text version of the example data
text<-"onset    terminus    tail    head
9/6/2000    9/7/2000    [email protected]  [email protected]
9/6/2000    9/7/2000    [email protected]    [email protected]
9/6/2000    9/7/2000    [email protected] [email protected]
9/6/2000    9/7/2000    [email protected]  [email protected]
5/7/2001    5/8/2001    [email protected]   [email protected]
9/6/2000    9/7/2000    [email protected] [email protected]
9/6/2000    9/7/2000    [email protected] [email protected]
9/6/2000    9/7/2000    [email protected]    [email protected]
9/6/2000    9/7/2000    [email protected]  [email protected]
9/6/2000    9/7/2000    [email protected] [email protected]
9/6/2000    9/7/2000    [email protected]   [email protected]
9/5/2000    9/6/2000    [email protected]   [email protected]
9/5/2000    9/6/2000    [email protected] [email protected]
9/5/2000    9/6/2000    [email protected] [email protected]
9/5/2000    9/6/2000    [email protected]  [email protected]
5/7/2001    5/8/2001    [email protected]   [email protected]
9/5/2000    9/6/2000    [email protected]   [email protected]
9/5/2000    9/6/2000    [email protected]   [email protected]
9/5/2000    9/6/2000    [email protected]    [email protected]
9/5/2000    9/6/2000    [email protected]  [email protected]
9/5/2000    9/6/2000    [email protected]   [email protected]
9/5/2000    9/6/2000    [email protected]    [email protected]
9/5/2000    9/6/2000    [email protected]    [email protected]
9/5/2000    9/6/2000    [email protected] [email protected]
8/31/2000   9/1/2000    [email protected]    [email protected]
8/31/2000   9/1/2000    [email protected]   [email protected]"

# write out the example data to an example input file
inputFile<-tempfile()
cat(text,file=inputFile)

现在，加载网络动态库

library(networkDynamic)

# read in tab-delimited example input file
timeData<-read.csv(inputFile,sep = "\t",stringsAsFactors = FALSE)
# check that it was loaded correctly
timeData

# convert the date formats into a numeric time (milliseconds)
timeData$onset<-as.numeric(as.POSIXct(timeData$onset,format='%m/%d/%Y'))
timeData$terminus<-as.numeric(as.POSIXct(timeData$terminus,format='%m/%d/%Y'))

# create a table of email address to map to numeric ids
emails<-unique(c(timeData$head,timeData$tail))

#covert ids
timeData$head<- match(timeData$head,emails)
timeData$tail<- match(timeData$tail,emails)

# convert to networkDynamic object
enronDyn<-networkDynamic(edge.spells=timeData)

# copy in the network names
network.vertex.names(enronDyn)<-emails

# load ndtv library
library(ndtv)

# compute the animation at 30-day interval
compute.animation(enronDyn,slice.par=list(start=967705200,end=989305200,interval=2592000,aggregate.dur=2592000,rule='latest'))
# render out the animation
render.animation(enronDyn)
ani.replay()

然而，你的输入数据对我来说看起来有点有趣。我很确定原始的安然电子邮件数据具有比电子邮件发送当天更精确的时间戳，并且发送每封电子邮件不应该花费一整天的时间？如果您可以找到具有更精确时间戳的数据版本，那么您在呈现和分析动态事件的方式上将具有更大的灵活性。例如，您将知道每天发送电子邮件的顺序等。