CODEWITHSUNDEEP
r
Fish
Fish

Reputation: 131

transform non-numeric data to numeric data with R

I have a csv file with this format :

android ; login.html , connect.json , page1.json 

windows ; login.html , connect.json , page1.json , page2.html , page5.html 

windows ; login.html , connect.json , page4.json

To do PCA multivariate analysis with these variables, these variable must be numeric like this :

1 ; 3  

0 ; 5

0 ; 3

0 or 1 to indicate whether windows or android followed by the number of pages. I am looking for a way to modify these non numeric data Any idea please? Best

Upvotes: 0

Views: 1092

Answers (2)

G. Grothendieck
G. Grothendieck

Reputation: 269501

Try strsplit and lengths:

DF <- read.table(text = Lines, sep = ";", as.is = TRUE, strip.white = TRUE)
transform(DF, V1 = as.numeric(V1 == "android"), V2 = lengths(strsplit(V2, ",")))

giving:

  V1 V2
1  1  3
2  0  5
3  0  3

Note: We used this input:

Lines <- "android ; login.html , connect.json , page1.json 
windows ; login.html , connect.json , page1.json , page2.html , page5.html 
windows ; login.html , connect.json , page4.json"

Upvotes: 1

A5C1D2H2I1M1N2O1R2T1
A5C1D2H2I1M1N2O1R2T1

Reputation: 193517

Here's one approach:

data.frame(V1 = as.numeric(mydf$V1 == "android"), 
           V2 = count.fields(textConnection(mydf$V2), sep = ","))
#   V1 V2
# 1  1  3
# 2  0  5
# 3  0  3

Sample data:

mydf <- read.table(
  header = FALSE, sep = ";", stringsAsFactors = FALSE, strip.white = TRUE,
  text = '"android" ; "login.html , connect.json , page1.json" 
"windows" ; "login.html , connect.json , page1.json , page2.html , page5.html" 
"windows" ; "login.html , connect.json , page4.json"')

Upvotes: 2

Related Questions