在 R 中创建个体寿命的相似性矩阵

Question

我认为这真的很简单，但是当我确定存在一个现成的功能时，一些谷歌搜索和翻阅 R 书并没有发现一个现成的功能来实现这一点。

我有一个包含 3 列的向量：个人标签、出生日期和死亡日期。我想创建一个方阵，其中每个单元格是两个人都活着的天数。

所以从这里：

ID Birth Death
A    1/5    5/5
B    2/5   30/5
C   10/5   31/5

到这里：

    A    B    C
A   4    3    0
B   3   28   20
C   0   20   21

什么是对角线（如此自重叠）在很大程度上无关紧要，因为在以后的分析中我可以忽略它。

为任何帮助干杯

Answer 1

所以一个朋友想出了这个办法，虽然有点长：

data<-read.csv("data.csv",header=TRUE) 
life_start<-as.Date(data$born,"%d/%m/%Y") # Get born column
life_finish<-as.Date(data$die,"%d/%m/%Y") # Get death column
lifespan<-as.numeric(life_finish-life_start) # Calculate lifespan
data<-cbind(data,lifespan) # Add lifespan to data
data
overlap_matrix<-matrix(vector(),length(data$ID),length(data$ID)) # Create empty matrix for overlap values
rownames(overlap_matrix)<-paste(data$ID) # Name rows with ID
colnames(overlap_matrix)<-paste(data$ID) # Name columns with ID

for (i in 1:length(data$ID)) { # Loop through every ID and call it i (the focal individual)

  born_a<-as.Date(data$born[i],"%d/%m/%Y") # Get birthday for focal individual
  die_a<-as.Date(data$die[i],"%d/%m/%Y") # Get deathday for focal individual
  lifespan_a<-data$lifespan[i] # Get lifespan for focal individual

  diag<-0+i # Calculate matrix ID of the diagonal value so the matrix isn't mirrored

  for (j in diag:length(data$ID)) { # Loop through every ID and call it j (to compare to i)

    born_b<-as.Date(data$born[j],"%d/%m/%Y") # Get birthday for comparison individual
    die_b<-as.Date(data$die[j],"%d/%m/%Y") # Get deathday for comparison individual
    lifespan_b<-data$lifespan[j] # Get lifespan for comparison individual

    if ((born_a <= die_b) && (die_a >= born_b)) { # If the focal individual was born before the comparison died, and died after the comparison was born (i.e. if there was overlap)

      born_diff<-as.numeric(born_a-born_b) # Calculate difference between birthdays
      die_diff<-as.numeric(die_a-die_b) #  Calculate difference between deathdays

      if(born_diff <= 0) { # If the focal individual was born before the comparison individual
        overlap<-lifespan_a+born_diff # Trim the lifespan of the focal indiviual by the difference (and add one to match days rather than the difference)
      } else {
        overlap<-lifespan_a # Start overlap as the lifespan of the focal individual
      }

      if(die_diff > 0) { # If focal individual died after the comparison individual
        overlap<-overlap-die_diff # Trim the lifespan of the focal individual by the difference (and add one to match the days rather than the difference)
      }

      overlap_matrix[i,j]<-overlap # Shunt overlap value to the overlap matrix

    } else { # No overlap!
      overlap_matrix[i,j]<-0 # Shunt the zero
    }

  }

}
overlap_matrix

“sna”包也有一个叫做“interval.graph”的函数来完成这项工作。