用R语言将数据转换成Format Vowpal Wabbit Input Files

相关文章：https://github.com/JohnLangford/vowpal_wabbit/wiki

https://github.com/JohnLangford/vowpal_wabbit/wiki/Input-format

# 从网上下载titanicDF的数据源

titanicDF<-read.csv(‘http://math.ucdenver.edu/RTutorial/titanic.txt‘,sep=‘\t‘)

#  将数据源存到本地

# write.table(titanicDF,‘titanicDF.txt‘,row.names=F)

# write.csv(titanicDF,"titanicDF.csv",row.names=F)

# 创建新变量 title

titanicDF$Title<-ifelse(grepl(‘Mr ‘,titanicDF$Name),‘Mr‘,ifelse(grepl(‘Mrs ‘,titanicDF$Name),‘Mrs‘,ifelse(grepl(‘Miss‘,titanicDF$Name),‘Miss‘,‘Nothing‘)))

# 转换成factor类型

titanicDF$Title<-as.factor(titanicDF$Title)

head(titanicDF)

##                                            Name PClass   Age    Sex

## 1                  Allen, Miss Elisabeth Walton    1st 29.00 female

## 2                   Allison, Miss Helen Loraine    1st  2.00 female

## 3           Allison, Mr Hudson Joshua Creighton    1st 30.00   male

## 4 Allison, Mrs Hudson JC (Bessie Waldo Daniels)    1st 25.00 female

## 5                 Allison, Master Hudson Trevor    1st  0.92   male

## 6                            Anderson, Mr Harry    1st 47.00   male

##   Survived   Title

## 1        1    Miss

## 2        0    Miss

## 3        0      Mr

## 4        0     Mrs

## 5        1 Nothing

## 6        1      Mr

str(titanicDF)

## ‘data.frame‘:    1313 obs. of  6 variables:

##  $ Name    : Factor w/ 1310 levels "Abbing, Mr Anthony",..: 22 25 26 27 24 31 45 46 50 54 ...

##  $ PClass  : Factor w/ 3 levels "1st","2nd","3rd": 1 1 1 1 1 1 1 1 1 1 ...

##  $ Age     : num  29 2 30 25 0.92 47 63 39 58 71 ...

##  $ Sex     : Factor w/ 2 levels "female","male": 1 1 2 1 2 2 1 2 1 2 ...

##  $ Survived: int  1 0 0 0 1 1 1 0 1 0 ...

##  $ Title   : Factor w/ 4 levels "Miss","Mr","Mrs",..: 1 1 2 3 4 2 1 2 3 2 ...

# 查看数据缺失情况

library(dfexplore)

## Loading required package: ggplot2

dfplot(titanicDF)

sum(is.na(titanicDF$Age)) # Age有557条记录数据缺失

## [1] 557

# 用中位数填补age变量的缺失数据

titanicDF$Age[is.na(titanicDF$Age)]<-median(titanicDF$Age,na.rm=T)

# 再次查看tittitanicDF数据是否有缺失值

dfplot(titanicDF)

# 对数据的列进行重组，将目标"Survived"放在最后面

titanicDF<-titanicDF[c("PClass","Age","Sex","Title","Survived")]

head(titanicDF)

##   PClass   Age    Sex   Title Survived

## 1    1st 29.00 female    Miss        1

## 2    1st  2.00 female    Miss        0

## 3    1st 30.00   male      Mr        0

## 4    1st 25.00 female     Mrs        0

## 5    1st  0.92   male Nothing        1

## 6    1st 47.00   male      Mr        1

# binarize all factors:对所有因子变量进行虚拟变量（哑变量）处理

library(caret)

## Loading required package: lattice

titanicDummy<-dummyVars(~.,data=titanicDF,fullRank=F)

titanicDF<-as.data.frame(predict(titanicDummy,newdata=titanicDF))

head(titanicDF)

##   PClass.1st PClass.2nd PClass.3rd   Age Sex.female Sex.male Title.Miss

## 1          1          0          0 29.00          1        0          1

## 2          1          0          0  2.00          1        0          1

## 3          1          0          0 30.00          0        1          0

## 4          1          0          0 25.00          1        0          0

## 5          1          0          0  0.92          0        1          0

## 6          1          0          0 47.00          0        1          0

##   Title.Mr Title.Mrs Title.Nothing Survived

## 1        0         0             0        1

## 2        0         0             0        0

## 3        1         0             0        0

## 4        0         1             0        0

## 5        0         0             1        1

## 6        1         0             0        1

https://github.com/JohnLangford/vowpal_wabbit/wiki/Input-format

[Label] [Importance [Tag]]|Namespace Features |Namespace Features … |Namespace Features

isclassification<-T

outcomeName<-"Survived"

labelName<-titanicDF[,"Survived"]

weightName<-""

objDF<-titanicDF

predictors<-names(objDF)[!names(objDF) %in% c(outcomeName,weightName)]

# LABELS & IMPORTANCE

if (is.null(labelName)) {

outcomeName<-"ignoreme"

objDF[,outcomeName]<-"0 |"

} else {

if (isclassification) {

# everything should be -1 1 for classification

objDF[,outcomeName]<-ifelse(objDF[,outcomeName]>0,1,-1)

}

if (weightName !=‘‘)

objDF[,outcomeName]<-psate(objDF[,outcomeName],objDF[,weightName],"|")

else

objDF[,outcomeName]<-paste(objDF[,outcomeName],"|")

}

# Pairing column names with data.. adding 1 blank character before each variable

for (i in predictors){

objDF[,i]<-ifelse(objDF[,i]==1,paste0(‘ ‘,i),

ifelse(objDF[,i]==0,‘‘,paste0(‘ ‘,i,":",objDF[,i])))

}

# reorder columns:重新组合列

objDF<-objDF[c(outcomeName,predictors)]

head(objDF)

##   Survived  PClass.1st PClass.2nd PClass.3rd       Age  Sex.female

## 1      1 |  PClass.1st                          Age:29  Sex.female

## 2     -1 |  PClass.1st                           Age:2  Sex.female

## 3     -1 |  PClass.1st                          Age:30

## 4     -1 |  PClass.1st                          Age:25  Sex.female

## 5      1 |  PClass.1st                        Age:0.92

## 6      1 |  PClass.1st                          Age:47

##    Sex.male  Title.Miss  Title.Mr  Title.Mrs  Title.Nothing

## 1            Title.Miss

## 2            Title.Miss

## 3  Sex.male              Title.Mr

## 4                                  Title.Mrs

## 5  Sex.male                                   Title.Nothing

## 6  Sex.male              Title.Mr

write.table(objDF,"vw.txt",sep="",quote=F,row.names=F,col.names=F)