About MiniRadde

MiniRadde · ‎08-22-2018

Thank you all for all the help you provided. The solutions seemed to be in a similar setting from all of you. I could not however make it work, as best it just copied the value from A or B but did not generate a sum. I solved the task by adding some data steps instead and could from there merge them back and make the sumation over variables instead. Ugly code - YES - but it solved my problem. Again, thanks for all the effort!

MiniRadde · ‎08-21-2018

Hi @novinosrin! yes, I do have it already inte the HAVE dataset. a smentioned there are other variables in there where TOT have values. They are however not sorted as illustrated.

MiniRadde · ‎08-21-2018

Dear all, I am in deep need of some help with something that I think is quite basic and simple, but which I cannot figure out myself. Suppose I have a dataset with a large number of variables and four of them are of concern for this data step: HAVE = Var1 Var2 Var3 Var4 A 1 C 1 B 1 C 2 TOT 1 C . A 2 C 4 B 2 C 5 TOT 2 C . A 3 C 7 B 3 C 8 TOT 3 C . A 1 D 90 B 1 D 80 TOT 1 D . A 2 D 60 B 2 D 50 TOT 2 D . A 3 D 30 B 3 D 20 TOT 3 D . Now I wish to fill out the missing values for the "TOT"-observation, by summing the observations of "A" and "B" in which Var2 and Var3 are equal. I.e., I need the red numbers in: WANT = Var1 Var2 Var3 Var4 A 1 C 1 B 1 C 2 TOT 1 C 3 A 2 C 4 B 2 C 5 TOT 2 C 9 A 3 C 7 B 3 C 8 TOT 3 C 15 A 1 D 90 B 1 D 80 TOT 1 D 170 A 2 D 60 B 2 D 50 TOT 2 D 110 A 3 D 30 B 3 D 20 TOT 3 D 50 Is there anyone who knows how to solve this in a simple data step? Thank you in advance!

MiniRadde · ‎09-01-2017

Thank you so much @Astounding, I tried to do this by several rounds of datasteps and sorting. This really really helöped me a lot and saved me (probably several days) of extra work! Thank you!

MiniRadde · ‎09-01-2017

I have a problem that I hope a friendly soul out there could help me to solve. I have a data set like DAY VALUE 1 1 1 2 1 3 1 6 2 8 2 6 2 4 3 15 3 10 3 35 3 20 3 5 and now I wish to do conditional operations so that for each DATE I create a new variable containing the average VALUE for that DAY, i.e. DAY VALUE Mean_PER_DAY 1 1 3 1 2 3 1 3 3 1 6 3 2 8 6 2 6 6 2 4 6 3 15 17 3 10 17 3 35 17 3 20 17 3 5 17 This is just a small subset of my dataset that contains thousands of DATE-observations, so using if DATE=2 etc. is not a good solution. Is there anyone who could write me some example code for this? Thank you in advance!

MiniRadde · ‎02-16-2017

Thank you so much @DanielSantos and @ballardw for the great effort. I chose to set all the datasets in a new one, although the file became enourmous, and then I used proc sort data= HAVE nodupkey out = WANT; by Var1 Var2 Var3... ; * All variables that must be equal if the observation should be deleted ; run; to delete the undesired observations. This felt most safe since I could understand everything in that code. Again, thank you for your effort.

MiniRadde · ‎02-16-2017

Again, thank you @Astounding for your help. That solution you proposed I already tried, and it does unfortuntly not work. I solved it very very badly but at least I can continue. Since all files have almost the same name (what differs is the date), I made 31 single conversion for each day in the month: data want_YYYY_MM_DD; set have._YYYY_MM_DD (rename=(TROUBLE_VARIABLE=TEMP)); TROUBLE_VARIABLE = put(TEMP, 7.); drop TEMP; run; Next by doing "find and replace" on the YYYY and MM numbers, I could repeat the procedure for every month and every year in my sample period. As said, this was very non-fancy, but it worked!

MiniRadde · ‎02-13-2017

Hi again, and sorry for bothering about this topic again. The solution worked fine as long as the dataset-formats were equal. When running over al my sets, I noticed that in some -not all- of my files there exists a variable X. Sometimes this one is defined as numeric and sometimes as character, resulting in an error: Variable X has been defined as both character and numeric I tried to solve it using set lib.data_2017: (rename=X) or (drop=X), and so on, but this won't work since the variable is not in all datasets. Any ideas on how to proceed?

MiniRadde · ‎02-09-2017

Thank you @PeterClemmensen. This solutions is however too simple since I have about 500 datasets, and I do not know which of them that are equal. They contain about 5000 observations each so I cannot check it either. What I am looking for is some commands that can perform something like for dataset 1 to 500 if dataset_i = dataset_: drop dataset end

MiniRadde · ‎02-09-2017

Hello. Is there somebody who knows how to delete a dataset from a library if another dataset has exactly the same content? I have a long, long, list of datasets representing each calender day of several years. At some dates the dataset itself is not updated, and I'd like to keep only those that tracks any change. The amount of duplicates may vary. Thank you in advance!

MiniRadde · ‎02-08-2017

Thank you @Astounding and @ChrisHemedinger, I am forever grateful to your quick support!

MiniRadde · ‎02-08-2017

Hi Astounding! Waow, I really did not expect help that fast! Your code works PERFEKT! Thank you so much for your help, I could not have figured this out myself in time!

MiniRadde · ‎02-08-2017

Hi ChrisHemedinger! Thank you so much for your help! The code worked almost perfect, the only thing was that it stored the entire file-name, but that is really a trifle. Thank you, and thank you again!

MiniRadde · ‎02-08-2017

Hello, I am stuck on a procedure that I really need to automize, but I am very new to writing own code in SAS. My problem could be summarized as follows: input > 500 datasets (named dataset_name) having X variables. output = 1 dataset having X+1 variables ("+1" = "name" in dataset_name). What this means is that I have a library that contains over 500 datasets that I'd like to consolidate. All files are named with the same prefix, followed by the date, e.g. data_2017-01-06 data_2017-01-18 data_2017-02-08, and so on. What I need to do is to create a new dataset containing all observations from all datasets, while I also need to be able to identify at which date the observation was made (i.e. the date-part of the dataset name). I would like to store only the YYYY-MM-DD part of the name as observations (preferably as characters). Is there any friendly soul out there who could help me to solve this problem?

Online Status	Offline
Date Last Visited	‎11-11-2019 08:37 AM

Re: Conditional summation of observations

Re: Conditional summation of observations

Conditional summation of observations

Re: Operations on unequally sized subsets of observations

Operations on unequally sized subsets of observations

Re: Delete Dataset if a Duplicate Exists

Re: Add dataset-name as new variable and repeat for a large number of ...

Re: Add dataset-name as new variable and repeat for a large number of ...

Re: Delete Dataset if a Duplicate Exists

Delete Dataset if a Duplicate Exists

Re: Delete Dataset if a Duplicate Exists

Re: Add dataset-name as new variable and repeat for a large number of ...

Re: Add dataset-name as new variable and repeat for a large number of ...

Re: Delete Dataset if a Duplicate Exists

Re: Add dataset-name as new variable and repeat for a large number of ...

Re: Conditional summation of observations

Re: Conditional summation of observations

Conditional summation of observations

Re: Operations on unequally sized subsets of observations

Operations on unequally sized subsets of observations

Re: Delete Dataset if a Duplicate Exists

Re: Add dataset-name as new variable and repeat for a large number of ...

Re: Add dataset-name as new variable and repeat for a large number of ...

Re: Delete Dataset if a Duplicate Exists

Delete Dataset if a Duplicate Exists

Re: Add dataset-name as new variable and repeat for a large number of ...

Re: Add dataset-name as new variable and repeat for a large number of ...

Re: Add dataset-name as new variable and repeat for a large number of ...

Add dataset-name as new variable and repeat for a large number of data...