title	author	date	output
r语言程序笔记	sjy	2023-09-26	html_document

# 常用r包 ```{r} library(openxlsx) #读写xlsx文件 library(tidyverse) #数据处理，绘图,批量化(map、pmap) library(modelr) #辅助建模(resample.crossv_kfold,seq_range,data_grid,add_prediction) library(broom) #模型信息提取(tidy,glance,augment) #统计建模包汇总 library(gee) #广义估计方程r包(不能做无序多分类logistic) library(VGAM) #vglm()gee无序多分类 library(MASS)#polr()有序logsitic回归 library(brant)#brant()检验有序logistic回归平行性 library(nnet)#multinom()无序多分类logsitic回归 library(survival) #coxph()cox回归建模 library(randomForestSRC) #rfsrc()随机生存森林 library(pROC) #roc()求roc所用数据 library(survivalROC) #survivalROC()求时间依赖roc所用数据 #统计绘图包 library(rms) #nomogram()列线图 library(riskRegression) #plotCalibration()校正曲线 library(pheatmap) #pheatmap()热图 library(RColorBrewer) #调色板 ``` # 使用across函数批量更改变量类型 ```{r} #批量设置因子型/有序型/字符型/数值型变量 variables % mutate(across(any_of(variables),as.character))#factor可替换为ordered/as.character/as.numeric #连续型变量转因子 #截断值用原数据集产生 #分位数切断 result1 %mutate(across(any_of(cols_to_cut), ~cut(., breaks = c(min(.,na.rm=T)-1,quantile(., probs = c(0.25, 0.5, 0.75,1),na.rm = TRUE)),labels=F), .names = "{.col}class4")) #极差等分 result3 % mutate(across(any_of(cols_to_cut), ~cut(., breaks=c(min(.,na.rm=T)-1,min(.,na.rm=T)+diff(range(.,na.rm=T))/4,min(.,na.rm=T)+2*diff(range(.,na.rm=T))/4,min(.,na.rm=T)+3*diff(range(.,na.rm=T))/4,max(.,na.rm=T)),labels=F), .names = "{.col}classd4")) #截断值用roc曲线求转二分类(截断值数据集thresholds见下文) #结合for循环 for (nm in predvars){ result2 % mutate(across( nm, ~cut(., breaks =c(min(.,na.rm=T)-1,thresholds[[match(nm, thresholds$predvar),1]],max(.,na.rm=T)),labels=F), .names = "{.col}class2"))} ``` # 批量计算新列 ```{r} #求日期间隔时间 for (i in 2:53){ diff_col_name 0.9){eGFR=141*(scr/0.9)**-1.209*0.993**年龄} else if(性别==0&scr<=0.9){eGFR=141*(scr/0.9)**-0.411*0.993**年龄} else if(性别==1&scr>0.7){eGFR=144*(scr/0.7)**-1.209*0.993**年龄} else if(性别==1&scr<=0.7){eGFR=144*(scr/0.7)**-0.329*0.993**年龄} else{eGFR=NA} return(eGFR) } } for(i in 1:53){ diff_col_name % select(sorted_colnames) #进阶方法2 num_at_end % mutate(fz2=case_when(str_detect(value,"Scr")~1, str_detect(value,"eGFR")~2, str_detect(value,"年龄")~3, str_detect(value,"时间")~4,)) zz % arrange(fz1, fz2) sorted_colnames % select(sorted_colnames) ``` # map批量建模 ```{r} #批量正态性检验 result %>% select(all_of(cols_to_cut)) %>% map(~shapiro.test(.)) variables % mutate(models=map(train,~coxph(Surv(a46,a45) ~ a3+a6+a7+a8+a9+a10+a12+a13+a16,.x))) %>% mutate(results=map(models,augment)) ```