cancel
Turn on suggestions
Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.
Showing results for 
Search instead for 
Did you mean: 
Kraussvan
‎04-05-2023
  • 5 Posts
  • 3 Likes Given
  • 0 Solutions
  • 0 Likes Received

Re: Stratified sampling

by Fluorite | Level 6 in Statistical Procedures
‎04-05-2023 12:20 PM
‎04-05-2023 12:20 PM
😁 Basically, you use a second step to create the 3 groups. You just don't do it using "surveyselect". My intention was, in the first "surveyselect", to create the group flag. Thank you anyway ... View more

Re: Stratified sampling

by Fluorite | Level 6 in Statistical Procedures
‎04-05-2023 10:07 AM
‎04-05-2023 10:07 AM
Hi Ballardw,   "Depends on your data. Are any Id values associated with different cluster values? Repeated within a cluster?" No, ID are key on the input data set.   "You might share the code you attempted. Best would be to include data that behaves like your working set."   Ksharp solution was pretty similar to my aproach.   "And yes you can select different sample sizes for different strata." Yes, the problem is to make all groups   "-The ID's are not replaceable (they are key in the first and last table)" Sorry, I mean the source data set and the final output.   Thank you for your time.   Best KV     ... View more

Re: Stratified sampling

by Fluorite | Level 6 in Statistical Procedures
‎04-05-2023 10:02 AM
‎04-05-2023 10:02 AM
Hi FreelanceReinh, This approach is much closer to what I was looking for. The "surveyselect" parameters already ensure non-replacement and division by stratified groups. The only improvement that could be made (I don't know if it's possible) is in the "surveyselect" that creates the 3 groups using a table with a greater number of records but it chooses only the number that is intended. (15 20 per group). Briefly, put together what the 1st and 2nd "surveyselect" does. Regardless of your answer, thank you for your time and availability and congratulations on the elegance of the solution. Best KV   ... View more

Re: Stratified sampling

by Fluorite | Level 6 in Statistical Procedures
‎04-05-2023 09:53 AM
‎04-05-2023 09:53 AM
Hi Ksharp, Thanks for your willingness to help. Your approach is to build blocks that eventually come together and use "surveyselect" only for random choice. My question is more about how to get the prerequisites compactly within the "surveyselect". But it works, no question about it. Thanks a lot for your time! ... View more

Stratified sampling

by Fluorite | Level 6 in Statistical Procedures
‎04-04-2023 06:57 PM
‎04-04-2023 06:57 PM
Hello,   I have what I think is a simple task, or, a set of simple tasks all together and I can't figure out how to do it. a) Imagine that I have one table with 2 columns: "ID", "cluster" (cluster={1,2}) with 200 rows. b) I want a new table with: "ID","cluster", "GROUP". that is composed by 3 samples of random rows, with "strata" by cluster,  BUT:             -I want different sizes for each cluster             -I want to determine that sizes (not with percentage)             -The ID's are not replaceable (they are key in the first and last table) In the end, I will get some counts like this:   GROUP:1     CLUSTER:1      Number of ID's: 15 GROUP:1     CLUSTER:2      Number of ID's: 25 GROUP:2     CLUSTER:1      Number of ID's: 15 GROUP:2     CLUSTER:2      Number of ID's: 25 GROUP:3     CLUSTER:1      Number of ID's: 15 GROUP:3     CLUSTER:2      Number of ID's: 25   The closer I got  was using the "proc surveyselect", but can't guarantee that ID's are not repeated and can't choose different sizes for different cluster. Can anyone out there help me, please?   Thank you in advance KV ... View more
Contact Me
Online Status
Offline
Date Last Visited
‎04-05-2023 04:01 PM
Likes given to
View All
Quick links