使用 grep 用 R 中的不同值替换每第 n 次和第 (n+1) 次出现

Question

For quarterly data

> df  
  TIME     GEO  Value
2000Q1 Austria 3864.6   
2000Q2 Austria 3841.3   
2000Q3 Austria 3843.0   
2000Q4 Austria 3847.2   
2001Q1 Austria 3853.5   
2001Q2 Austria 3875.2   
2001Q3 Austria 3886.7  
2001Q4 Austria 3921.9   
2002Q1 Austria 3865.2   
2002Q2 Austria 3872.4  
2002Q3 Austria 3876.0  
2002Q4 Austria 3887.9   
2003Q1 Austria 3938.3   
2003Q2 Austria 3954.5  
2003Q3 Austria 3972.8  
2003Q4 Austria 3971.9  

I'm naively converting quarterly data to monthly with df.mon <- rep(df$Value, each=3) . I do the same for df$TIME

 df.mon$TIME <- rep(df$TIME, each=3)  

I want to convert these time identifiers to monthly ones so that I can easily use df.mon as a weight on monthly data.

So, I have

  >head(df.mon, n=10)
     GEO  month
  3864.6 2000Q1
  3864.6 2000Q1
  3864.6 2000Q1
  3841.3 2000Q2
  3841.3 2000Q2
  3841.3 2000Q2
  3843.0 2000Q3
  3843.0 2000Q3
  3843.0 2000Q3
  3847.2 2000Q4

I want replace the 1st, 4th, 7th, etc. occurrence of Q1 with M01 , and the 2nd, 5th, 8th, etc. occurence of Q1 with M02 and so on, to produce:

     GEO  month
  3864.6 2000M01
  3864.6 2000M02
  3864.6 2000M03
  3841.3 2000M04
  3841.3 2000M05
  3841.3 2000M06
  3843.0 2000M07

The closest explanation to this is here , and seems like using grep with back references \\1 is the way to go (a useful list is here ).

I've tried,

gsub("(?:Q1)", "\\1M01\\2M02\\3M03", df.mon$month)

which only gives me

     2000M01M02M03
     2000M01M02M03
     2000M01M02M03

I've tried other specifications like gsub("(?:Q1)(?:Q1)(?:Q1)", "\\\\1M01\\\\2M02\\\\3M03", df.mon$month) , for which no replacements are made.

I don't really understand what's going on with the (?: ) command (and it seems unnecessary), and I don't know Perl so I'm at a loss for how to make this seemingly easy replacement work.

Answer 1

Try

year <- grep("[0-9]{4}", df.mon$month, value=T)
month <- paste("M", 1:12, sep="")
yearmonth <- paste(year, month, sep="")
df.mon$month <- yearmonth

No complicated reg expressions needed.