Hi, I have two datasets:  ds1 Advisor JP_Morgan JP_Morgan_Chase___Co JP_Morgan_Secs__Asia JP_Morgan_Securities Lehman_Brothers Lehman_Brothers_Kuhn Laidlaw___Co__UK__Lt Laidlaw_Adams___Peck Laidlaw_Ansbacher Laidlaw_Equities_Inc Macquarie_Capital__U Macquarie_Capital_Ad Macquarie_Capital_Gr Macquarie_Capital_Pa   ds2 Advisor JP_Morgan Laidlaw Macquarie_Capital Lehman_Brothers     When I run the code below using soundex and compged ( I set a threshold level of 100 for the compged) I get the outcome given below the code:  data advisors; length possible_match $3; set ds1; tmp1=soundex(advisor); possible_match='No'; do i=1 to nobs; set ds2(rename=(advisor=advisor2)) point=i nobs=nobs; tmp2=soundex(advisor2); dif=compged(tmp1,tmp2); if dif<=100 then do; possible_match='Yes'; drop i tmp1 tmp2; output; end; end; if possible_match='No' then do; call missing(advisor2); output; end; run; Output: possible_match Advisor advisor2 dif Yes JP_Morgan JP_Morgan 0 No JP_Morgan_Chase___Co   590 No JP_Morgan_Secs__Asia   590 No JP_Morgan_Securities   690 Yes Lehman_Brothers Lehman_Brothers 0 Yes Lehman_Brothers_Kuhn Lehman_Brothers 70 No Laidlaw___Co__UK__Lt   610 No Laidlaw_Adams___Peck   600 No Laidlaw_Ansbacher   520 No Laidlaw_Equities_Inc   600 Yes Macquarie_Capital__U Macquarie_Capital 0 Yes Macquarie_Capital_Ad Macquarie_Capital 50 Yes Macquarie_Capital_Gr Macquarie_Capital 100 Yes Macquarie_Capital_Pa Macquarie_Capital 50   But when I change the threshold score to a different value, say 500 then the compged scores (given by the diff column) changes entirely! Why is this? How can I prevent this from happening?   Also in the third column, rather than getting blanks ( if the compegd score is greater than the threshold) if I want to get the same name in the second column how should the code be amended?   Any help would be greatly appreciated! 🙂   Thanks!     ... View more

Hi, I have two datasets:  ds1 Advisor JP_Morgan JP_Morgan_Chase___Co JP_Morgan_Secs__Asia JP_Morgan_Securities Lehman_Brothers Lehman_Brothers_Kuhn Laidlaw___Co__UK__Lt Laidlaw_Adams___Peck Laidlaw_Ansbacher Laidlaw_Equities_Inc Macquarie_Capital__U Macquarie_Capital_Ad Macquarie_Capital_Gr Macquarie_Capital_Pa   ds2 Advisor JP_Morgan Laidlaw Macquarie_Capital Lehman_Brothers       When I run the code below using soundex and compged ( I set a threshold level of 100 for the compged) I get the outcome given below the code:    data advisors; set ds1; tmp1=soundex(advisor); do i=1 to nobs; set ds2(rename=(advisor=advisor2)) point=i nobs=nobs; tmp2=soundex(advisor2); dif=compged(tmp1,tmp2); if dif<=100 then do; possible_match='Yes'; drop i tmp1 tmp2; output; end; end; run;   Advisor advisor2 dif possible_match JP_Morgan JP_Morgan 0 Yes Lehman_Brothers Lehman_Brothers 0 Yes Lehman_Brothers_Kuhn Lehman_Brothers 70 Yes Macquarie_Capital__U Macquarie_Capital 0 Yes Macquarie_Capital_Ad Macquarie_Capital 50 Yes Macquarie_Capital_Gr Macquarie_Capital 100 Yes Macquarie_Capital_Pa Macquarie_Capital 50 Yes   However, say I want an output table which gives me all the 14 observations in ds1, gives 'yes' for possible match and the respective match from ds2 in the second column if the compegd is below the threshold and gives 'no' for possible match and a blank for the second column if the threshold is surpassed.    Basically I want to know how each observation in ds1 fairs with the matching and not to lose the observations because there wasn't a close match.    Thank you! ... View more

Hi, I have a timeseries dataset. In it, I have a variable that contains company names. But the company names change in many observations across time. I want to match and replace the company names so that companies  which are (obviously) the same have the same name in the dataset and not several different names.   Ex: XYZ,XYZ Company, XYZ & CO, XYZ CO etc. all will have one name such as for example 'XYZ Company'. ... View more