R diff() 和 0 值
我想使用 diff(log(myvar)) 来计算一些时间序列日志返回。在这种情况下，myvar 中的某些值设置为 0，并且 diff() 返回 Inf，因为它应该在数学上。
当滞后对的任一数据点为 0 时，如何强制 diff() 发生异常并返回 0？例如，
diff(log(c(0,1,2,3,4)))
将返回
0 0.6931472 0.4054651 0.2876821
强制第一个计算值为 0 而不是 Inf。
更新 我实际上在使用 is.infinite() 时遇到了问题。考虑以下情况：
> v = diff(log(c(1, 0, 0, 2, 3)))
> v
[1] -Inf NaN Inf 0.4054651
> is.infinite(v)
[1] TRUE FALSE TRUE FALSE
这里有 3 种不同的情况：-Inf、Inf 和 NaN。目标是将所有基本报价为 0 的回报设置为 0，因此我必须添加 v[is.nan(v)] = 0。
为什么我要将收益设置为 0 而不是 NA？嗯，这更多的是一个逻辑问题，而不是一个编程问题，但我的想法是，在我的例子中，0 意味着参考价格没有改变（并不是说它是未知的）。
I want to use diff(log(myvar)) to calculate some time-series log returns. Some of the values in myvar are set to 0 and diff() returns Inf in this case, as it should mathematically.
How can I force diff() to make an exception and return 0 when either data point of the lagged pair is 0? For example
diff(log(c(0,1,2,3,4)))
Would return
0 0.6931472 0.4054651 0.2876821
Forcing the very first calculated value to be 0 rather than Inf.
UPDATE
I actually had a problem using is.infinite(). Consider the following case:
> v = diff(log(c(1, 0, 0, 2, 3)))
> v
[1] -Inf NaN Inf 0.4054651
> is.infinite(v)
[1] TRUE FALSE TRUE FALSE
Here we have 3 different cases, -Inf, Inf and NaN. The goal is set 0 to all returns whose base quotation is 0 so I had to add v[is.nan(v)] = 0.
Why would I want to set returns to 0 rather than NA? Well this is more of a logical problem rather than a programming question, but the idea is that in my case 0 means that the reference price hasn't changed (and not that it is unknown).
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。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（3）
你也可以这样做：
You could also do this:
有什么原因吗
这个愚蠢的解决方案不起作用 ？ （假设您的数据一开始只有非有限值，我认为获得
Inf或-Inf的唯一方法是具有零值.. .)Is there a reason the boneheaded solution of
doesn't work? (Assuming you only have non-finite values in your data to begin with, I think the only way you can get
Infor-Infis to have a zero value ...)还应该添加
0/0 情况。
Should also add
for 0/0 case.