qq_43872984的博客

数据格式：Excel中的格式：导入r后的格式：setwd(mypath)myfolder<-dir()mypattern<-"[:alpha:]*(\():alpha:)"i=1for (name_folder in myfolder) {myfile<-paste(mypath,“3-0.csv”,sep="/")#print(myfile)tmp<-read.csv(myfile) %>% select_if(~ !all(is.na(.)))tmp&

@mydata %>% select_if(~ !all(is.na(.)))mydata %>% select_if(~ !any(is.na(.)))

https://mlr3gallery.mlr-org.com/posts/2020-03-11-basics-german-credit/备查：mlr3计算features importance#classiftask = TaskClassif$new("GermanCredit", german, target = "credit_risk")train_set = sample(task$row_ids, 0.8 * task$nrow)test_set = setdif.

非常好的mlr3案例集合，原文来源于：https://mlr3gallery.mlr-org.com/posts/2020-05-04-moneyball/这里仅仅是把代码复制下来方便以后的查找和使用：library("mlr3")library("mlr3learners")library("mlr3pipelines")requireNamespace("mlr3measures")library("mlr3data")# 查看缺失数据skim(moneyball)#

mycols<-c("red1","blue","purple","darkblue","darkgreen","orange","gold","brown","tomato","skyblue","turquoise","salmon")p <- ggplot(mtcars, aes(mpg, wt)) + geom_point(aes(colour = factor(cyl)))p + scale_colour_manual(values = mycols)[1] ...

常用的数据分列函数是separate，相当好用，另外一函数是extract:library(tidyr)df<- data.frame(x=c("nlvt35","pulvt33"))pattern<- "([:alpha:]{1})([:alpha:]{1,6})([:digit:]{1,3})"df %>% extract(x,into=c("p","devicetype","vdd"),regex=pattern)result: p devicetyp..

经常遇到计算fail rate的事情，案例如下：myfun<-function(x) (length(x[x>0.3])/length(x))iris.grouped %>% summarise_if(is.numeric, myfun)

有时候数据结构如下：前三行均表示某个测试点，需要对数据做长宽变化lot kt6666 kt6666 kt6666 kt6666 kt6666 kt6666 kt6666 kt6666 kt6666 wafer 6 6 6 6 6 6 6 6 6 site 1 2 3 4 5 6 7 8 9 vt1 0.098437 0.

ggplot()+geom_point(data=iris,aes(Sepal.Width,Sepal.Length,size=1.2,shape=Species,colour=Species))+scale_colour_manual(values=myp[1:3])+geom_point(data=tt,aes(Sepal.Width,Sepal.Length),color="black",size=3)不成功的做法，ggplot(iris,aes(Sepal.Width,Sepal.Lengt

居然浪费我·一晚上的时间newdata<-ddply(mtcars,.(cyl),summarize,median=median(mpg))ggplot(mtcars,aes(cyl,mpg))+geom_boxplot()+geom_text(data=newdata,aes(cyl,median,label=median)max.mpg<-max(mtcars$mpg) ggplot(mtcars,aes(factor(cyl),mpg))+geom_boxplot()+geom

通常需要在圆上划分成若干的小方块，然后标注图形或者数字，这里是用R实现的代码：radius<-30.48a<-0b<-0angle<-0:360*(pi/180)x<-a+radius*cos(angle)/2y<-b+radius*sin(angle)/2plot(x,y,type="p",cex=0.5,col="black",asp=1)n...

交互式包plotly在R完全和ggplot2兼容，只需ggplotly(p+…),即使用自带的plotly函数也很方便，语法规则和ggplot2很相似，仿照help文件轻松搞定。

R包的另类安装：把另外一台电脑装好的同版本的R下的lib文件夹压缩打包解压缩到当前电脑R下的lib文件夹，即可。另外在有网络的条件下，需要安装R包还是通过R自带的install package方便，比如装tidyverse，选中安装，所有辅助包一起安装。...

多写函数，充分利用R中的循环语句tidyverse是功能包合集，强大fun.outlier可以更加简单的，应该利用boxplot.stats(x)$out来写fun.outlier<- function(x,time.iqr=1.5) { outlier.low<-quantile(x,probs=c(0.25))-IQR(x)*time.iqr outlier.high&l...