ilikesas Tracker

random numbers without repetition

ilikesas — Tue, 12 Oct 2021 23:05:52 GMT

Hi,

I want to generate say 100 data points each having 4 variables, and each variable is a random number between 1 and 10 such that each of the 4 variables is unique for a given data point, i.e, no repetition on the data points level.

Thank you.

forecasting with categorical variables

ilikesas — Fri, 29 May 2020 17:13:29 GMT

Hi,

I have a time series with the following independent variables:

time, vendor, product, miles driven

time is the time variable (1,2,3,... corresponding to months)

vendor and product are categorical variables

miles driven is a quantitative variable

Will it be statistically correct if I did the following:

proc autoreg data=MyData;
 class vendor product;
model sales = vendor product miles_driven;
 run;

Its just that the few times when I did time series I only had time as an independent variable, but here there are categorical and quantitative variables as well so I just want to make sure that using the proc autoreg will yield meaningful results.

Thank you

forecasting with categorical and quantitative variables

ilikesas — Thu, 28 May 2020 20:40:09 GMT

Hi,

I have a time series with the following independent variables:

time, vendor, product, miles driven

time is the time variable

vendor and product are categorical variables

miles driven is a quantitative variable

Will it be statistically correct if I did the following:

proc autoreg data=MyData;
 class vendor product;
model sales = vendor product miles_driven;
 run;

Thank you

accessing variable names through loop

ilikesas — Sun, 11 Aug 2019 22:47:16 GMT

I have the following:

data have;
input num x1 x2 x3;
datalines;
1 . . .
2 . . .
3 . . .
;
run;

suppose i want to make a multiplication table (pseudo code)

data want;

set have;

do i=1 to 3

x&i = number*i;

end

run;

Basically, the variables x1-x3 contain in them information, namely, the multiplication factor, so I want using a loop to "assemble" the variable name.

Thank you

Re: Obtain all possible pairs from a list

ilikesas — Fri, 15 Jun 2018 01:36:13 GMT

These combinations include duplicates and same-element combinations, whereas I don't want these duplicates and same element combinations - how would I in this case delete the duplicates and same value combs?

Re: Obtain all possible pairs from a list

ilikesas — Thu, 14 Jun 2018 22:33:28 GMT

Hi PaigeMiller,

this code is quite simple, and I didn't even think of it!

But this code has repetitions, that is, since I have 5 elements it gives me all the pairs twice and also the pair of each element with itself

5sq = 25 pairs,

but if I want the unique non-same pairs I should get (25-5)/2 = 10 pairs

Thanks!

Re: Obtain all possible pairs from a list

ilikesas — Thu, 14 Jun 2018 22:30:12 GMT

Hi novinsorin,

I am not familiar with hash, is it possible to get the solution in a macro form similar to this:

proc sql;
select element
into :element_list separated by ' '
from have
;
quit;
proc sql;
select count(element)
into :n
from have
;
quit;

%put &element_list;
%put &n;

%macro m1;
data want;
%do i=1 %to &n;
%let var1=%scan(&element_list,&i);
var1=&var1;

%end;
%mend m1;
%m1

Thank you

Obtain all possible pairs from a list

ilikesas — Thu, 14 Jun 2018 22:07:59 GMT

Hi,

I have the following data:

data have;
input element$ 1-2 qt;
datalines;
A 1
B 2
Cc3
D 4
Ee5
;
run;

I want to obtain all possible pairs like this:

A B

A Cc

A D

A Ee

B Cc

B D

...

Thanks!

output regression statistic into a table

ilikesas — Mon, 22 Jan 2018 01:07:55 GMT

Hi,

I am trying to do the following:

proc reg data=Sashelp.Cars plots=none;  
   model Horsepower = EngineSize Weight / vif r influence;
output out=the_output
rstudent = rstudent
h =h
cookd = cookd;
run;

basically I would like to have the statistic such as rstudent and cookd to be output into a new table, the_output, but it is always empty.

Thank you

Re: selecting observations based on their relative position to other observations

ilikesas — Sun, 21 Jan 2018 17:34:30 GMT

Hi,

as you can see I included the additional "industry" variable.

So I guess that for more variables I will have to repeat the process.

Thanks!

Re: selecting observations based on their relative position to other observations

ilikesas — Sun, 14 Jan 2018 00:45:28 GMT

Thanks for the code.

I managed to include an additional "industry" variable:

data want;
set have2 ;
retain _flag 0 _line _n _line2 _n2 ;
if find(line, "stock")>0 then do;
__n=0;
_flag=1;
_n=n;
_line=line;
end;
if _flag then __n+1;
if __n=3 then do;
_n2=n;
_line2=line;
end;
if __n=5 then do;
n=_n;
industry=line;
price=_line2;
line=_line;
output;
end;
drop _: ;
run;

so I guess that if I had more variables to create I would just repeat the process

Re: selecting observations based on their relative position to other observations

ilikesas — Mon, 08 Jan 2018 03:06:48 GMT

Hi novinosrin,

for the line in your code:

else if index(line,'$')>0 then do;

if I understand correctly the code looks for the symbol "$", but in my case I have many useless lines that contain the symbol "$", so it is really the relative position of the lines from the line that contains the word "stock" that are important

Re: selecting observations based on their relative position to other observations

ilikesas — Sun, 07 Jan 2018 17:19:24 GMT

Hi draycut,

thanks for the code, worked perfectly!

If possible to expand the code, in my case I have more variables of interest. For example, the 4th line after the line that contains the word "stock" I have its industry:

data have2;
input n line $50.;
datalines;
30  doidn
31  fifeniofe
32  doewiij
33  fofugrugr
34	text jdjdjf stock IBM
35	djdff 
36	25$
37  fg fgg
38	computers
39	text jdjdjf stock AAA
40	djdff 
41	22$
42  fggh
43	food
44	text jdjdjf stock BBB
45	djdff 
46	55$
47  duehuefhu
48	automotive
49  dfeoufeijer
50  ierhde43848
;
run;

Using your code I did the following:

data want2;
   merge have2 have2(firstobs=3 rename=(line=price)) have2(firstobs=5 rename=(line=industry));
   if find(line, "stock")>0 then output;
run;

And also got the result that I wanted!

I would like to ask for your opinion on the following:

1) In my data I have tens to hundreds of thousands of lines (most of them useless) and 5 to 7 variables that I need to extract, so this would mean merging 5 to 7 times the same data table just like above. Is this a good approach, i.e, could there be efficiency or some other problems (like the computer freezing for example)?

2) Recently I worked on some data where I also had to restructure it into a more analysis-friendly form using Excel VBA. Looking at my data I would perhaps do the following:

Loop through my original data until I find the word "stock". Once I find it, remember the cells row r and paste the cell's content into

a new spreadsheet in cell (1,2) (the first column in the new sheet is the stock, second column price etc).

In the original sheet the contents of the cell with row = r+2 are the stock's price and so they will be pasted into the new sheet's cell (2,2), and so on for the other variables...

And when I reach a new cell with the word "stock" I go to the next line in the new sheet.

It seems, at least to me, that using cells is easier conceptually (because I couldn't figure out how to do the above mentioned data restructuring using SAS), so is it possible to "cellulize" data in SAS? Is it even worth it?

Thanks

Re: selecting observations based on their relative position to other observations

ilikesas — Sun, 07 Jan 2018 16:22:05 GMT

Hi mkeintz,

since my data is generated externally, lets assume that I already imported data equivalent to:

data have;
input line $ 1-10;  
datalines;
xxx
stock AAA
aaa
11$
xxx
xxx
stock BBBB
bbb
22$
run;

It is now that I need to do the process described in the question. I tried to guess based on your code by doing the following:

data want;
set have;
input @ 'stock' stock :$5. // price :$5.;
put (_all_) (=);

run;

But unsurprisingly it didn't work

selecting observations based on their relative position to other observations

ilikesas — Sun, 07 Jan 2018 16:07:02 GMT

Hi,

suppose I have a table with a somewhat complex data structure like the following:

n	line
…	…
34	text jdjdjf stock IBM
35	djdff
36	25$
…	…

Whenever I have a line with the word "stock" in it, the actual price is always 2 lines below. So what I would like to have is the following structure:

n	line	price
34	text jdjdjf stock IBM	25$

Like this the stock and its price are on the same observations and its possible to make an analysis.

Thank you.

Re: Fixed effects with interactions

ilikesas — Wed, 03 Jan 2018 21:24:47 GMT

The paper upon which we base our research indeed treats the variables as fixed effects and even mentions that he used SAS proc glm, so the project supervisor wants to use fixed effects as well. But indeed the variables somewhat look like random effects, so she said that we will also do a mixed model!

Re: Fixed effects with interactions

ilikesas — Wed, 03 Jan 2018 21:01:27 GMT

I hope that I am not becoming too annoying...

when I do model Cholesterol = Sex BP_Status weight smoking / ss3; I do get only the SS3, but now I don't have the coefficients of the independent variables and their t values. Is it possible to get them without the coefficients of the categories of the CLASS variables? My concern is that if I have too many variables due to the fixed effects SAS will get overloaded.

Re: Fixed effects with interactions

ilikesas — Wed, 03 Jan 2018 16:34:04 GMT

Hi Rick,

I tried to run the following code:

proc glm data=sashelp.heart;
   class sex BP_Status;
  model Cholesterol = Sex BP_Status weight smoking / solution ;
   quit;

Is it possible to get the overall significance of each class variable (i.e, Type III SS) without outputting the actual statistics of each category of each class variable, because I have hundreds of CEOs (and of other categorical variables which I might add to the model).

Thanks!

Re: Fixed effects with interactions

ilikesas — Mon, 18 Dec 2017 15:09:25 GMT

I am still fairly new to the concepts of fixed, random, mixed effects (and the SAS applications of), but I know that in my regression each CEO is a dummy variable because each CEO as an individual is different and therefore each CEO's effect is different when controlled for other variables.

Could I use the code that you gave me before:

/* same analysis by using the CLASS statement */
proc glm data=have;
   class CEO;              /* generates dummy variables internally */
   model return = CEO market_cap debt CEO*market_cap / solution;
   ods select ParameterEstimates;
quit;

Thanks!

PS: do you have any articles about fixed,random, mixed?

Re: regression with 2 dummy types

ilikesas — Mon, 18 Dec 2017 00:51:31 GMT

Hi Rick,

I posted a new question https://communities.sas.com/t5/SAS-Statistical-Procedures/Tip-Fixed-vs-Random-Effects-in-Panel-Data/td-p/421894

it looks very similar to the question that you answered here, but could you please take a look at it, since I am not sure.

Thanks!