library(twang)
star<- stardata
names(star)
summary(star)

#recode gender
star$Gender[star$Gender=="M"] <- 1
star$Gender[star$Gender=="F"] <- 2
#recode race
star$Race[star$Race=="White"] <- 1
star$Race[star$Race=="Black"] <- 2
star$Race[star$Race=="Asian"] <- 3
star$Race[star$Race=="Native American"] <- 4
star$Race[star$Race=="Pacific Islander"] <- 5
star$Race[star$Race=="Multi-racial"] <- 6

#removing ID column
star$`Deidentified Patient ID` <- NULL

#omit null values
star1<- na.omit(star)

#write as csv
write.csv(star1, file = "star_new.csv")

#OMIT X1 column 
star_new$`X1` <- NULL

star2<-star_new
View(star2)
star2$gender.f <- factor(star2$Gender)
star2$race.f <- factor(star2$Race)
names(star2)

#recode the variable names
star2$gen_anxiety_disorder<-star2$`Generalized Anxiety Diagnosis`
star2$years_schooling_completed <- star2$`Years of schooling completed`
star2$age_at_first_MDE <- star2$`Age at First MDE`
star2$years_since_first_MDE <- star2$`Years Since First MDE`
star2$ months_current_episode <- star2$`Months in current episode`
star2$is_recurrent <- star2$`Is Recurrent`
star2$family_depression <-star2$`Family Hx Depression`

star2$family_bipolar<-star2$`Family Hx Bipolar`
star2$panic<-star2$`Panic Diagnosis`
star2$social_phobia<-star2$`Social Phobia Diagnosis`
star2$OCD <-star2$`Obsessive Compulsive Diagnosis`
star2$alcohol_abuse<-star2$`Alcohol Abuse Diagnosis`
star2$drug_abuse <-star2$`Drug Abuse Diagnosis`
star2$qids_score<-star2$`Drug Abuse Diagnosis`
star2$hrsd_score<-star2$`HRSD-17 (ROA)`
star2$cit_success<-star2$`CIT Success`
###################################################################