[R] help with text patterns in strings

[arun]

HI Burnette,
As this is continuation of the earlier thread, you could post it on the same thread by cc: to rhelp.


Try this:
res1<-sapply(vec3,function(x) length(vec2New[grep(x,vec2New)]) )
dat1<-data.frame(res1,Name=names(vec3))

 dat1$Name<-factor(dat1$Name,levels=c("early","mid","late","wknd"))
 with(dat1,tapply(res1,list(Name),FUN=sum))
#early   mid  late  wknd 
 #   0     1     4     6 

#or
 sapply(split(res1,names(vec3)),sum)
#early  late   mid  wknd 
 #   0     4     1     6 
A.K.

----- Original Message -----
From: "Crombie, Burnette N" <bcrombie at utk.edu>
To: arun <smartpink111 at yahoo.com>
Cc: 
Sent: Wednesday, June 19, 2013 3:55 PM
Subject: RE: [R] help with text patterns in strings

Arun, let me know if I should post this email separately, but it involves the script from our previous conversation.  I've been messing around as I think of potential scenarios with my data and am unclear how I can recount vec3 after assigning range names to the different days of the week.  For this example, I want my output to go from:
#Su  M Tu  W Th  F Sa
# 2    0   0     1   1    3  4
to:
# early   mid   late   wknd
#   0          1          4          6

Thanks for your help throughout, but, again, let me know if I should start a new thread.

Burnette

##########################################
Begin script
##########################################
dat3<- read.csv("~/Rburnette/TextStringMatch.csv", stringsAsFactors=FALSE)
dat3
#respondent.ID            response
# 1                       Friday
# 2                       Wednesday
# 3                       Friday, saturday,Sunday
# 4                       Saturday
# 5                       Sat, sun
# 6                       Th,F, Sa

# Rename the variable “response” to “Ans” to fit the script that’s already been written
# fix(dat3) can be used to do this manually, but then you need to keep "dat3" as the data frame, not "dat3edit"
# if not familiar, fix() generates a popup window like a spreadsheet that can be edited, and character vs numeric property can be changed
# the data set being “fixed” is saved automatically upon closing, but I think only within the current R session
# I think you need to redefine the fix() as a new object to keep the changes outside the R session  (need to test this)
##########################################
library(gdata)
dat3edit <- rename.vars(dat3,from="response", to="Ans")
dat3edit
#respondent.ID            Ans
# 1                       Friday
# 2                       Wednesday
# 3                       Friday, saturday,Sunday
# 4                       Saturday
# 5                       Sat, sun
# 6                       Th,F, Sa

# get rid of the spaces embedded in text strings
##########################################
dat3edit$Ans2 <- gsub(" ","",dat3edit$Ans)
dat3edit$Ans2
# [1] "Friday"                 "Wednesday"              "Friday,saturday,Sunday" "Saturday"              
# [5] "Sat,sun"                "Th,F,Sa" 

# split up multiple responses within an observation so they can be counted separately
##########################################
vec2<-unlist(strsplit(dat3edit$Ans2,","))
vec2
# [1] "Friday"    "Wednesday" "Friday"    "saturday"  "Sunday"    "Saturday"  "Sat"       "sun"      
# [9] "Th"        "F"         "Sa" 

#consistently format all (split up) responses to start with a capital letter for more accurate matching to a “universal” response code created in the next step
##########################################
library(Hmisc)
vec2New<-capitalize(vec2)
vec2New
# [1] "Friday"    "Wednesday" "Friday"    "Saturday"  "Sunday"    "Saturday"  "Sat"       "Sun"      
# [9] "Th"        "F"         "Sa"

#match capitalized data to a “universal” response code of choice
##########################################
vec3<- c("Su","M","Tu","W","Th","F","Sa")
sapply(vec3,function(x) length(vec2New[grep(x,vec2New)]) )
#Su  M Tu  W Th  F Sa
# 2  0  0  1  1  3  4

#assign range names to vec3
##########################################
names(vec3) <- c("wknd","early","early","mid","late","late","wknd")
vec3
# wknd early early   mid  late  late  wknd 
# "Su"   "M"  "Tu"   "W"  "Th"   "F"  "Sa"