Changing string variable with a comma as a decimal separator to numeric variable

[tig som]

My variable «mystring» is a string variable that uses commas as decimal separator
Below is its list
----------------------- copy starting from the next line -----------------------

* Example generated by -dataex-. For more info, type help dataex
clear
input str6 mystring
"8,5"  
"'"    
""     
"8"    
"4"    
"0,4"  
"12,896"
end

------------------ copy up to and including the previous line ------------------

Listed 7 out of 7 observations

I want to change this variable to numeric variables
I used the command below.
destring mystring, replace force ignore(",")
the results are

dataex

----------------------- copy starting from the next line -----------------------

* Example generated by -dataex-. For more info, type help dataex
clear
input int mystring
   85
    .
    .
    8
    4
    4
12896
end

------------------ copy up to and including the previous line ------------------

Listed 7 out of 7 observations


As you can see, the new numbers are the wrong number. For example, the first variable was supposed to be 9 or 8.5, but now it is written as 85 instead. Also, the last variable is supposed to be 12.896 or 13 ( if rounded). However, now I have a very big value, which is equal to 12896

Would you please give me a suggestion on how I can change this string variable to numeric while keeping the right amount of values?

Thank you, and looking forward to hearing from you

[Clyde Schechter]

-destring- has a -dpcomma- option that will cause it to interpret the comma as the decimal point. Use that. When you -ignore(",")-, you cause -destring- to behave as if the commas just aren't there--so that 8,5 becomes 85.

By the way, in your example data, the second observation has "'" in the second observation. That is neither a missing value nor a valid number. So -destring- will tell you that it can't proceed because of that. If your real data set contains problems like that, you either need to run some commands to -drop- such observations or -replace- their values with missings or legal numbers. Alternatively, -ignore("'")- would tell -destring- to act is if there aren't any 's in the data--which will solve the problem if stray 's are the only non-number entries.

[tig som]

Thank you very much, Clyde Schechter!!