PST
包是否始终显示从右到左的上下文?
在query()
函数中,我们使用字符串来表示上下文。如果我假设从右到左指定了上下文(因为它似乎在print()
和cmine()
函数中),并且我对序列A->B->C
感兴趣,然后我应该查询:
query(S1.p1, "C-B-A")
此外,在predict()
函数中,我们使用seqdef()
来定义要预测的序列。这是否意味着我应该从左到右指定它们,因为TraMineR通常会这样做?
x <- seqdef("A-B-C)
predict(S1.p1, x)
答案 0 :(得分:1)
在概率后缀树(PST)中,当我们从根开始读取时,分支从右到左定义后缀。在第一级,您有后缀的最后一个元素,在级别2,您有最后一个元素之前的元素,等等。打印的树显示为左侧的根,并从左向右展开。然而,应该从左到右自然地读取在打印结果的节点中显示的后缀。例如,节点a-b-c
表示末尾带有c
的后缀。通过在左侧添加b-c
,从节点a
获得这样的节点。
同样适用于cmine
的结果。对于每个找到的上下文, a-b-c
,cmine
给出了在上下文之后立即获取每个可能状态的概率,即在示例中的c
之后。
总之,序列和上下文总是从左到右显示,即使上下文是从右到左构建的。
因此,如果您想查询序列A->B->C
,请使用query(S1.p1, "A-B-C")
。同样,要使用predict
预测特定序列,请从左到右自然定义序列。
答案 1 :(得分:1)
应从左到右阅读序列。以下代码提供了对此的验证:
library(PST)
data.seq <- seqdef("A-B-C-D-E-F")
S1.test <- pstree(data.seq, ymin = 0.001, lik = FALSE, with.missing = FALSE)
print(S1.test)
--(e)-[ p=(0.2,0.2,0.2,0.2,0.2,0.2) - n=6 ]
`--(A)-[ p=(0.001,0.995,0.001,0.001,0.001,0.001) - n=1 ]--|
`--(B)-[ p=(0.001,0.001,0.995,0.001,0.001,0.001) - n=1 ]
`--(A-B)-[ p=(0.001,0.001,0.995,0.001,0.001,0.001) - n=1 ]--|
`--(C)-[ p=(0.001,0.001,0.001,0.995,0.001,0.001) - n=1 ]
`--(B-C)-[ p=(0.001,0.001,0.001,0.995,0.001,0.001) - n=1 ]
`--(A-B-C)-[ p=(0.001,0.001,0.001,0.995,0.001,0.001) - n=1 ]--|
`--(D)-[ p=(0.001,0.001,0.001,0.001,0.995,0.001) - n=1 ]
`--(C-D)-[ p=(0.001,0.001,0.001,0.001,0.995,0.001) - n=1 ]
`--(B-C-D)-[ p=(0.001,0.001,0.001,0.001,0.995,0.001) - n=1 ]
`--(A-B-C-D)-[ p=(0.001,0.001,0.001,0.001,0.995,0.001) - n=1 ]--|
`--(E)-[ p=(0.001,0.001,0.001,0.001,0.001,0.995) - n=1 ]
`--(D-E)-[ p=(0.001,0.001,0.001,0.001,0.001,0.995) - n=1 ]
`--(C-D-E)-[ p=(0.001,0.001,0.001,0.001,0.001,0.995) - n=1 ]
`--(B-C-D-E)-[ p=(0.001,0.001,0.001,0.001,0.001,0.995) - n=1 ]
`--(A-B-C-D-E)-[ p=(0.001,0.001,0.001,0.001,0.001,0.995) - n=1 ]--|
plot(S1.test)
它还表明应该从底部到顶部读取绘制的树。