Percent Change for Multiple Variables
Locked
 
|
** Dear SPSSers,
The following was adapted using syntax developed by Art Kendal. It is very useful for quickly computing percent change and overall change between the earliest year and the most recent year for one variable. What I was wondering is can the syntax be further refined to compute multiple percentage changes for each of the variable without having to repeat the syntax and rename the target variables in each case? The syntax would compute percent change for each variable and then the "aggfile" would compute the total changes for each variable as well - adding the variable name in each row. Please let me know if you have any questions concerning my question and thank you in advance for your assistance. Sincerely Damir * ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// . DATA LIST LIST / YEAR (F4) TOTALARREST (F6) PART1 (F6) PART2 (F6) POPULATION (F6) TOTALARRESTRATE PART1_R PART2_R . BEGIN DATA 1998 54573 10909 43664 1020521 5347.6 1069.0 4278.6 1999 63322 10555 52767 1042196 6075.8 1012.8 5063.1 2000 68363 9810 58553 1131184 6043.5 867.2 5176.3 2001 71553 11096 60457 1154464 6197.9 961.1 5236.8 2002 69409 11262 58147 1183197 5866.2 951.8 4914.4 2003 72381 10939 61442 1211448 5974.8 903.0 5071.8 2004 71054 10525 60529 1242270 5719.7 847.2 4872.5 2005 75031 10409 64622 1265900 5927.1 822.3 5104.8 2006 73424 10504 62920 1287987 5700.7 815.5 4885.1 2007 81743 11676 70067 1295033 6312.0 901.6 5410.4 2008 85981 12902 73079 1294654 6641.2 996.6 5644.7 2009 74337 12723 61614 1287344 5774.4 988.3 4786.1 2010 70479 11146 59333 1286461 5478.5 866.4 4612.1 2011 62530 10774 51756 1325758 4716.5 812.7 3903.9 2012 60625 10880 49745 1335415 4539.8 814.7 3725.1 END DATA . DATASET NAME CRIME . COMPUTE arrestchange=( totalarrest / lag(totalarrest) - 1 )*100. FORMATS arrestchange (pct9.2). VARIABLE LABELS ARRESTCHANGE 'year to year pct change'. LIST . COMPUTE nobreak = 1. DATASET DECLARE aggfile. AGGREGATE OUTFILE=aggfile /break = nobreak /entries = n /start = first(totalarrest) /end = last (totalarrest). DATASET ACTIVATE aggfile. COMPUTE periods= entries-1. COMPUTE grosschange= ( (end /start)-1) * 100. COMPUTE avgchange = grosschange/periods. COMPUTE growthrate =((end/start) **(1/periods))-1. COMPUTE annualgrowthpct = 100*growthrate. FORMATS grosschange avgchange annualgrowthpct (pct7.2) growthrate (f8.5). COMPUTE check = start*((1+growthrate)**periods). FORMATS check (f6). LIST . |
Re: Percent Change for Multiple Variables
Administrator
|
See VARSTOCASES command (INDEX subcommand) and BREAK subcommand in AGGREGATE.
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me. --- "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" |
Re: Percent Change for Multiple Variables
|
The original posting doesn't seem to have hit SPSSX-L yet; come on, Nabble!
Anyway, at 10:33 AM 2/12/2014, David Marso quoted DKUKEC: >The following was adapted using syntax developed by Art Kendal. It >is very useful for quickly computing percent change and overall >change between the earliest year and the most recent year for one >variable. Can the syntax be further refined to compute multiple >percentage changes for each of the variable without having to repeat >the syntax and rename the target variables in each case? First, are you thinking of this for year-to-year change? >COMPUTE arrestchange=( totalarrest / lag(totalarrest) - 1 )*100. >FORMATS arrestchange (pct9.2). >VARIABLE LABELS ARRESTCHANGE 'year to year pct change'. You don't nearly have to repeat the whole thing. Making a guess what variables you want to compute change for, you could write (code not tested), DO REPEAT Value = TOTALARREST PART1 PART2 POPULATION /Change= ARRESTCHANGE PT1CH PT2CH POPCHANGE. . COMPUTE Change =( Value / lag(Value) - 1 )*100. END REPEAT. FORMATS ARRESTCHANGE PT1CH PT2CH POPCHANGE (pct9.2). For the change from earliest to latest values, you can name any number of summary variables on the AGGREGATE: AGGREGATE OUTFILE=aggfile /break = nobreak /entries = n /A_Arrest = first(totalarrest) /Z_Arrest = last (totalarrest) /A_Part1 = first(PART1) /Z_Part1 = last (PART1) ... Then, in the aggregated dataset (DATASET ACTIVATE aggfile) you can define and compute the 'change' variable for each of your underlying variables; again, DO REPEAT can keep you from having to repeat the code again and again. Alternatively, you could do what David Marso suggested, and 'unroll' your data so each variable is in a separate group of cases. Then the calculations need only be done once. However, re-assembling into 'wide' form can be a bear. (Code, as always, is in an APPENDIX at the end of this posting.) |-----------------------------|---------------------------| |Output Created |12-FEB-2014 17:55:57 | |-----------------------------|---------------------------| [CRIME] YEAR TOTALARREST PART1 PART2 POPULATION TOTALARRESTRATE PART1_R PART2_R 1998 54573 10909 43664 1E+006 5347.60 1069.00 4278.60 1999 63322 10555 52767 1E+006 6075.80 1012.80 5063.10 2000 68363 9810 58553 1E+006 6043.50 867.20 5176.30 2001 71553 11096 60457 1E+006 6197.90 961.10 5236.80 2002 69409 11262 58147 1E+006 5866.20 951.80 4914.40 2003 72381 10939 61442 1E+006 5974.80 903.00 5071.80 2004 71054 10525 60529 1E+006 5719.70 847.20 4872.50 2005 75031 10409 64622 1E+006 5927.10 822.30 5104.80 2006 73424 10504 62920 1E+006 5700.70 815.50 4885.10 2007 81743 11676 70067 1E+006 6312.00 901.60 5410.40 2008 85981 12902 73079 1E+006 6641.20 996.60 5644.70 2009 74337 12723 61614 1E+006 5774.40 988.30 4786.10 2010 70479 11146 59333 1E+006 5478.50 866.40 4612.10 2011 62530 10774 51756 1E+006 4716.50 812.70 3903.90 2012 60625 10880 49745 1E+006 4539.80 814.70 3725.10 Number of cases read: 15 Number of cases listed: 15 DATASET COPY Unroll. DATASET ACTIVATE Unroll. VARSTOCASES /ID = id /MAKE Value FROM TOTALARREST PART1 PART2 POPULATION TOTALARRESTRATE PART1_R PART2_R /INDEX = Variable(Value) /KEEP = YEAR /NULL = KEEP. Variables to Cases |-----------------------------|---------------------------| |Output Created |12-FEB-2014 17:55:57 | |-----------------------------|---------------------------| [Unroll] Generated Variables |--------|------| |Name |Label | |--------|------| |id |<none>| |Variable|<none>| |Value |<none>| |--------|------| Processing Statistics |-------------|-| |Variables In |8| |Variables Out|4| |-------------|-| SORT CASES BY Variable YEAR. FORMATS Value (F9.2). LIST. List |-----------------------------|---------------------------| |Output Created |12-FEB-2014 17:55:57 | |-----------------------------|---------------------------| [Unroll] id YEAR Variable Value 1 1998 PART1 10909.00 2 1999 PART1 10555.00 3 2000 PART1 9810.00 4 2001 PART1 11096.00 5 2002 PART1 11262.00 6 2003 PART1 10939.00 7 2004 PART1 10525.00 8 2005 PART1 10409.00 9 2006 PART1 10504.00 10 2007 PART1 11676.00 11 2008 PART1 12902.00 12 2009 PART1 12723.00 13 2010 PART1 11146.00 14 2011 PART1 10774.00 15 2012 PART1 10880.00 1 1998 PART1_R 1069.00 2 1999 PART1_R 1012.80 3 2000 PART1_R 867.20 4 2001 PART1_R 961.10 5 2002 PART1_R 951.80 6 2003 PART1_R 903.00 7 2004 PART1_R 847.20 8 2005 PART1_R 822.30 9 2006 PART1_R 815.50 10 2007 PART1_R 901.60 11 2008 PART1_R 996.60 12 2009 PART1_R 988.30 13 2010 PART1_R 866.40 14 2011 PART1_R 812.70 15 2012 PART1_R 814.70 1 1998 PART2 43664.00 2 1999 PART2 52767.00 ... Number of cases read: 105 Number of cases listed: 105 ============================= APPENDIX: Test data, and code (not saved separately) ============================= DATA LIST LIST / YEAR (F4) TOTALARREST (F6) PART1 (F6) PART2 (F6) POPULATION (F6) TOTALARRESTRATE PART1_R PART2_R . BEGIN DATA 1998 54573 10909 43664 1020521 5347.6 1069.0 4278.6 1999 63322 10555 52767 1042196 6075.8 1012.8 5063.1 2000 68363 9810 58553 1131184 6043.5 867.2 5176.3 2001 71553 11096 60457 1154464 6197.9 961.1 5236.8 2002 69409 11262 58147 1183197 5866.2 951.8 4914.4 2003 72381 10939 61442 1211448 5974.8 903.0 5071.8 2004 71054 10525 60529 1242270 5719.7 847.2 4872.5 2005 75031 10409 64622 1265900 5927.1 822.3 5104.8 2006 73424 10504 62920 1287987 5700.7 815.5 4885.1 2007 81743 11676 70067 1295033 6312.0 901.6 5410.4 2008 85981 12902 73079 1294654 6641.2 996.6 5644.7 2009 74337 12723 61614 1287344 5774.4 988.3 4786.1 2010 70479 11146 59333 1286461 5478.5 866.4 4612.1 2011 62530 10774 51756 1325758 4716.5 812.7 3903.9 2012 60625 10880 49745 1335415 4539.8 814.7 3725.1 END DATA . DATASET NAME CRIME . LIST. DATASET ACTIVATE Crime DATASET COPY Unroll. DATASET ACTIVATE Unroll. VARSTOCASES /ID = id /MAKE Value FROM TOTALARREST PART1 PART2 POPULATION TOTALARRESTRATE PART1_R PART2_R /INDEX = Variable(Value) /KEEP = YEAR /NULL = KEEP. SORT CASES BY Variable YEAR. FORMATS Value (F9.2). LIST. ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |