当前位置:网站首页>decast id.var measure.var数据拆分与合并
decast id.var measure.var数据拆分与合并
2022-04-23 12:58:00 【qq_52813185】
R之data.table -melt/dcast(数据拆分和合并)
写在前面:数据整形的过程确实和揉面团有些类似,先将数据通过melt()函数将数据揉开,然后再通过dcast()函数将数据重塑成想要的形状
reshape2包:
melt-把宽格式数据转化成长格式。
cast-把长格式数据转化成宽格式。(dcast-输出时返回一个数据框。acast-输出时返回一个向量/矩阵/数组。)
注:melt是数据融合的意思,它做的工作其实就是把数据由“宽”转“长”。
cast 函数的作用除了还原数据外,还可以对数据进行整合。
dcast 输出数据框。公式的左边每个变量都会作为结果中的一列,而右边的变量被当成因子类型,每个水平都会在结果中产生一列。
tidyr包:
gather-把宽度较大的数据转换成一个更长的形式,它类比于从reshape2包中融合函数的功能
spread-把长的数据转换成一个更宽的形式,它类比于从reshape2包中铸造函数的功能。
data.table包:
data.table的函数melt 和dcast 是增强包reshape2里同名函数的扩展
library(data.table)
ID <- c(NA,1,2,2)
Time <- c(1,2,NA,1)
X1 <- c(5,3,NA,2)
X2 <- c(NA,5,1,4)
mydata <- data.table(ID,Time,X1,X2)
mydata
## ID Time X1 X2
## 1: NA 1 5 NA
## 2: 1 2 3 5
## 3: 2 NA NA 1
## 4: 2 1 2 4
md <- melt(mydata, id=c("ID","Time")) #or md <- melt(mydata, id=1:2)
#melt以使每一行都是一个唯一的标识符-变量组合
md #将第一列作为id列,其他列全部融合就可以了
## ID Time variable value
## 1: NA 1 X1 5
## 2: 1 2 X1 3
## 3: 2 NA X1 NA
## 4: 2 1 X1 2
## 5: NA 1 X2 NA
## 6: 1 2 X2 5
## 7: 2 NA X2 1
## 8: 2 1 X2 4
将变量"variable",和"value"揉合在一起,结果产生了新的两列,一列是变量variable,指代是哪个揉合变量,另外一列是取值value,即变量对应的值。我们也称这样逐行排列的方式称为长数据格式
melt:数据集的融合是将它重构为这样一种格式:每个测量变量独占一行,行中带有要唯一确定这个测量所需的标识符变量。
str(mydata)
## Classes 'data.table' and 'data.frame': 4 obs. of 4 variables:
## $ ID : num NA 1 2 2
## $ Time: num 1 2 NA 1
## $ X1 : num 5 3 NA 2
## $ X2 : num NA 5 1 4
## - attr(*, ".internal.selfref")=<externalptr>
str(md)
## Classes 'data.table' and 'data.frame': 8 obs. of 4 variables:
## $ ID : num NA 1 2 2 NA 1 2 2
## $ Time : num 1 2 NA 1 1 2 NA 1
## $ variable: Factor w/ 2 levels "X1","X2": 1 1 1 1 2 2 2 2
## $ value : num 5 3 NA 2 NA 5 1 4
## - attr(*, ".internal.selfref")=<externalptr>
setcolorder(md,c("ID","variable","Time","value")) ##setcolorder()可以用来修改列的顺序。
md
## ID variable Time value
## 1: NA X1 1 5
## 2: 1 X1 2 3
## 3: 2 X1 NA NA
## 4: 2 X1 1 2
## 5: NA X2 1 NA
## 6: 1 X2 2 5
## 7: 2 X2 NA 1
## 8: 2 X2 1 4
mdr <- melt(mydata, id=c("ID","Time"),variable.name="Xzl",value.name="Vzl",na.rm = TRUE) #variable.name定义变量名
mdr
## ID Time Xzl Vzl
## 1: NA 1 X1 5
## 2: 1 2 X1 3
## 3: 2 1 X1 2
## 4: 1 2 X2 5
## 5: 2 NA X2 1
## 6: 2 1 X2 4
mdr1 <- melt(mydata, id=c("ID","Time"),variable.name="Xzl",value.name="Vzl",measure.vars=c("X1"),na.rm = TRUE) #measure.vars筛选
mdr1
## ID Time Xzl Vzl
## 1: NA 1 X1 5
## 2: 1 2 X1 3
## 3: 2 1 X1 2
md[Time==1]
## ID variable Time value
## 1: NA X1 1 5
## 2: 2 X1 1 2
## 3: NA X2 1 NA
## 4: 2 X2 1 4
md[Time==2]
## ID variable Time value
## 1: 1 X1 2 3
## 2: 1 X2 2 5
#执行整合
# rowvar1 + rowvar2 + ... ~ colvar1 + colvar2 + ...
# 在这个公式中,rowvar1 + rowvar2 + ... 定义了要划掉的变量集合,以确定各行的内容,而colvar1 + colvar2 + ... 则定义了要划掉的、确定各列内容的变量集合。
newmd<- dcast(md, ID~variable, mean)
newmd
## ID X1 X2
## 1: 1 3 5.0
## 2: 2 NA 2.5
## 3: NA 5 NA
newmd2<- dcast(md, ID+variable~Time)
newmd2
## ID variable 1 2 NA
## 1: 1 X1 NA 3 NA
## 2: 1 X2 NA 5 NA
## 3: 2 X1 2 NA NA
## 4: 2 X2 4 NA 1
## 5: NA X1 5 NA NA
## 6: NA X2 NA NA NA
#ID+variable~Time 使用Time对(ID,variable)分组 Time:1,2,NA 类似excel的数据透析
newmd3<- dcast(md, ID~variable+Time)
newmd3 #variable:X1,X2 Time:1,2,NA 类似excel的数据透析
## ID X1_1 X1_2 X1_NA X2_1 X2_2 X2_NA
## 1: 1 NA 3 NA NA 5 NA
## 2: 2 2 NA NA 4 NA 1
## 3: NA 5 NA NA NA NA NA
即使只是凡世中一颗小小的尘埃,命运也要由自己主宰,像向日葵般,迎向阳光、勇敢盛开
版权声明
本文为[qq_52813185]所创,转载请带上原文链接,感谢
https://blog.csdn.net/qq_52813185/article/details/124359601
边栏推荐
- STM32 is connected to the motor drive, the DuPont line supplies power, and then the back burning problem
- There is no need to crack the markdown editing tool typora
- Subscribe to Alibaba demo send business messages
- Pytorch: a pit about the implementation of gradreverselayer
- SSM framework series - annotation development day2-2
- leetcode:437. 路径总和 III【dfs 选还是不选?】
- 只是不断地建构平台,不断地收拢流量,并不能够做好产业互联网
- Remote sensing image classification and recognition system based on convolutional neural network
- 进程虚拟地址空间区域划分
- 梳理網絡IP代理的幾大用途
猜你喜欢
Huawei cloud MVP email
Image attribute of input: type attribute of fashion cloud learning -h5
Remote sensing image classification and recognition system based on convolutional neural network
云原生KubeSphere部署Mysql
【csnote】ER图
【蓝桥杯】4月17日省赛刷题训练(前3道题)
Record the problems encountered in using v-print
SSM框架系列——Junit单元测试优化day2-3
Record some NPM related problems (messy records)
世界读书日:我想推荐这几本书
随机推荐
Read the data in Presto through sparksql and save it to Clickhouse
Importerror after tensorflow installation: DLL load failed: the specified module cannot be found, and the domestic installation is slow
Redis deployment of cloud native kubesphere
Common problems of unity (1)
21 days learning mongodb notes
31. 下一个排列
拥抱机器视觉新蓝海,冀为好望开启数字经济发展新“冀”遇
使用Source Insight查看编辑源代码
Record a website for querying compatibility, string Replaceall() compatibility error
将新增和编辑的数据同步更新到列表
Packet capturing and sorting -- TCP protocol [8]
Image attribute of input: type attribute of fashion cloud learning -h5
SSL certificate refund instructions
There is no need to crack the markdown editing tool typora
Bert base Chinese Download (SMART)
如何实现点击一下物体播放一次动画
Customize classloader and implement hot deployment - use loadclass
梳理網絡IP代理的幾大用途
BaseRecyclerViewAdapterHelper 实现下拉刷新和上拉加载
Go iris framework implements multi service Demo: start (listen to port 8084) service 2 through the interface in service 1 (listen to port 8083)