topic Re: creating ranges and then percentages in Statistical Procedures

creating ranges and then percentages

raqthesolid — Mon, 28 Sep 2020 19:43:26 GMT

Hello Folks,
I have a data with Firm, Year, and ROA variables.
MY ROA variables ranges from -50% to +50%.
I want to create ranges using ROA which looks like

Range Percentage

less than 0 percentage of firms in this category

0 to1% percentage of firms in this category

2 to 3% percentage of firms in this category

4 to 5% percentage of firms in this category

......50% percentage of firms in this category

Data sample:

Firm Year ROA

PK0000201019	2016	11.70255
PK0000301017	2016	9.428861
PK0000401015	2016	-5.80082
PK0000601010	2016	17.6943
PK0001101010	2016	-8.47214
PK0001301016	2016	20.18228
PK0001901013	2016	11.68819
PK0002001011	2016	13.31986
PK0002101019	2016	2.865007
PK0002201017	2016	9.238791
PK0002301015	2016	19.7908
PK0002501010	2016	1.970355
PK0002701016	2016	64.1466
PK0002901012	2016	12.86187
PK0003001010	2016	-3.51228
PK0003101018	2016	19.38653
PK0003201016	2016	3.417498
PK0003301014	2016	22.99802
PK0003401012	2016	-10.3915
PK0003801013	2016	15.53839
PK0003901011	2016	4.804186
PK0004101017	2016	7.54267
PK0004301013	2016	22.41655
PK0004401011	2016	9.554636
PK0004701014	2016	8.438554
PK0004801012	2016	8.95593
PK0004901010	2016	7.656038
PK0005201014	2016	50.03889
PK0005401010	2016	-14.8514
PK0005501017	2016	8.194166
PK0005701013	2016	25.27504
PK0006001017	2016	3.329774
PK0006101015	2016	5.472684
PK0006801010	2016	10.02528
PK0006901018	2016	9.398788
PK0007101014	2016	29.06179
PK0008101013	2016	5.884868
PK0008201011	2016	18.7473
PK0008401017	2016	14.02341
PK0008501014	2016	7.427927
PK0009201010	2016	21.76193
PK0009401016	2016	-7.2675
PK0009501013	2016	-0.98955
PK0009801017	2016	30.41894
PK0010001011	2016	5.148667
PK0010301015	2016	-2.02785
PK0010501010	2016	14.67657
PK0010901012	2016	12.50834
PK0011001010	2016	6.203303
PK0011101018	2016	16.16196
PK0011201016	2016	22.07683
PK0011301014	2016	33.96201
PK0011501019	2016	-6.85826
PK0011701015	2016	34.67594
PK0012101017	2016	12.91066
PK0012401011	2016	10.27558
PK0012501018	2016	14.37119
PK0013401010	2016	4.465276
PK0014201013	2016	9.261005
PK0014501016	2016	9.95184

Thanks for any help.

Re: creating ranges and then percentages

SylvainTremblay — Mon, 28 Sep 2020 19:57:20 GMT

Hello,

A simple strategy here would be to create and use User Defined Formats in SAS, to create your ranges.

Here is a reference:

Building and Using User Defined Formats

https://support.sas.com/resources/papers/proceedings/proceedings/sugi29/236-29.pdf

Sylvain

Re: creating ranges and then percentages

Reeza — Mon, 28 Sep 2020 20:03:59 GMT

What happened to 1 to 2%?

I would use FLOOR()/CEIL() to floor the variable to the nearest integer and use that since your ranges are 1's essentially. If your ranges encompass more than one number you could either use MOD() to account for that or you could create custom formats.

Or just apply a format to it.

proc freq data=have;
table roa;
format roa 8.;
run;

Re: creating ranges and then percentages

StatDave — Tue, 29 Sep 2020 13:48:00 GMT

Any time that you want to assign observations into quantile groups, of any consistent width, the easiest way to do that is with the GROUPS= option in PROC RANK. By specifying GROUPS=x, you create a variable with value 0, 1, 2, ... , x indicating which x-tile each observation is in. So, GROUPS=4 creates a variable indicating which quartile each observation is in. In your case, if you want an indicator of each percentile, specify GROUPS=100. For example, these statements create data set MYRANKS which is a copy of you input data set and includes the variable ROApctl with values 0,1, 2, 3, ... , 99 which indicates the percentile range for each observation.

proc rank groups=100 out=myranks;

var ROA; ranks ROApctl;

run;