第4R

#数据操作
hyper <-read.csv(‘http://archive.ics.uci.edu/ml/machine-learning-databases/thyroid-disease/hypothyroid.data’,
header=F)
names <- read.csv(‘http://archive.ics.uci.edu/ml/machine-learning-databases/thyroid-disease/hypothyroid.names’,
header=F, sep=’\t’)[[1]]
hyper
names
#替换
names <-gsub(pattern =":", replacement="", x = names)
#加表头
colnames(hyper)<-names
#显示表头
str(colnames(hyper))
#改名字
colnames(hyper)[1]<-“target”
colnames (hyper)
head(hyper)
#判断并替换
hyper$target<-ifelse(hyper$target==“negative”, 0, 1)
head(hyper)
#检查 0 1
table(hyperKaTeX parse error: Expected 'EOF', got '#' at position 9: target) #̲计算相对丰度 prop.tab…target))

#将变量target变成因子型
str(hyper)
hyper$target<-as.factor(hyper$target)
str(hyper)
#if(!require (DMWR))
#install.packages(“DMwR”)
#library(DMwR)
hyper_new<-SMOTE(target~.,hyper,perc.over = 100,
perc.under = 400)
table(hyper_new$target)

user<- read.csv(“E://dasan//数据挖掘//精准医疗大赛-初赛(复赛)数据//f_train_20180204.csv”)
#加表头
#colnames(user)<-names
table(user$label)user$label<-as.factor(user$label)hype{r}_{n}ew<-SMOTE(label.,user,perc.over=100,perc.under=200)table(hype{r}_{n}ew$label)

install.packages(“caret”)
library(“caret”)

#数据抽样 查看分类占比
x<-seq(1,10)
a<-sample(x,8,replace=FALSE)
(c<-sample(x,15,xreplace=T))
user<-read.csv(“data1.csv”,T)
colnames (user)<-c()
nrow(user)

sample小例子

set.seed(1234)

创建对象x，有1~10组成

x <- seq(1,10);x
typeof(x)
str(x)

利用sample函数对x进行无放回抽样

a <- sample(x,8,replace=FALSE);a

利用sample函数对x进行有放回抽样

b <- sample(x,8,replace=TRUE);b

当size大于x的长度

(c <- sample(x,15,replace = F))
(c <- sample(x,15,replace = T))

利用sample对活跃用户数据进行抽样

导入数据

#user <- read.csv(“活跃用户是否付费数据.csv”,T)
user <- read.csv(“data1.csv”,F)
user_col_names <- c(“用户id”,“是否付费”,“注册至今距离天数”,“最后一周登陆天数”,“最后一周登陆次数”,“最后一周0-8点登陆次数”,“最后一周8-18点登陆次数”,“最后一周18-24点登陆次数”)
colnames(user) <- user_col_names

查看数据user的行数

nrow(user)

利用sample函数对user数据进行无放回抽样

#set.seed(1234)

提取下标集

index <- sample(nrow(user),10000,replace=F)
head(index)

将抽样数据赋予对象user_sample

user_sample <- user[index,]

查看user_sample的行数

nrow(user_sample)

现在我们分别查看user与user_sample变量“是否付费”中0、1占比。

table(user$是否付费)table(use{r}_{s}ample$是否付费)
round(prop.table(table(user$是否付费)),3)round(prop.table(table(use{r}_{s}ample$是否付费)),3)
rate <- sum(user$是否付费==0)/nrow(user)d<-1:nrow(user)user[d[user$是否付费0],]#下标
user[userKaTeX parse error: Expected 'EOF', got '#' at position 10: 是否付费==0,]#̲booeal index1 <…是否付费0], 10000*rate)
index2 <-sample(d[user$是否付费==1],10000\ast (1-rate))use{r}_{s}amplel<-user[c(index1,index2),]tail(use{r}_{s}amplel$是否付费)

user <- read.csv(“E://dasan//数据挖掘//精准医疗大赛-初赛(复赛)数据//f_train_20180204.csv”,T)
index <- sample(nrow(user),500,replace=F)
head(index)
user_sample <- user[index,]i
nrow(user_sample)
table(user$SNP1)table(use{r}_{s}ample$SNP1)
round(prop.table(table(user$SNP1)),2)rate<-round(prop.table(table(use{r}_{s}ample$SNP1)),2)
#rate <- sum(user$SNP1==1)/nrow(user)d<-1:nrow(user)user[user$SNP13,2]
index1 <-sample(d[user$SNP1==1],500\ast rate[[1]])index2<-sample(d[user$SNP12], 500*rate[[2]])
index3 <-sample(d[user$SNP1==3],500\ast rate[[3]])use{r}_{s}amplel<-user[c(index1,index2,index3),]head(use{r}_{s}amplel$SNP1)
round(prop.table(table(user_samplel$SNP1)),2)

#d<-1:nrow(user)
#index1 <-d[user$SNP1==1]index1<-sample(d[user$SNP11],226)
index2 <-sample(d[user$SNP1==2],226)index3<-sample(d[user$SNP13],226)
length(index3)
user_samplel <- user[c(index1,index2,index3),]
round(prop.table(table(user_samplel$SNP1)),2)

round(prop.table(table(user_samplel$SNP1)),5)table(use{r}_{s}amplel$SNP1)
splitindex2 <- createDataPartition (iris$Species,times=2,p=0.1,list=TRUE)user$是否付费 <- as.factor(user$是否付费)

提取下标集

ind <- createDataPartition(user$是否付费,p=10000/nrow(user),
times=1,list=FALSE)

查看子集中0、1占比

prop.table(table(user[ind,‘是否付费’]))