About piecze

piecze · ‎06-22-2021

Hi all! I am trying to score customer's basket from the ASSOCS dataset with proc astore. The binary scoring model has been generated using Market Basket Analysis task in SAS Studio, which has also provided me with 130 association rules. I have then created a sample input dataset containing 2 customers (each with 7 products copied from the original dataset) to run through the scoring model on SAS Viya. However, I am struggling to intepret the results shown in the images below. The scoring model seems to list all indexes of the association rules as columns and assign the "rule_id" column values from rules set in to them. The values are arranged in increasing order and at some point switch to 0's only (image 2). What I would expect is simply a row with cutomer_id and a single rule_id or item with highest probabilty of occuring among the items in the basket. Any help on intepretation or how to make use of the scoring model would be highly appreciated! 🙂

piecze · ‎10-21-2020

It worked! Thank you so much @Sam_SAS! I actually wanted to use the calculated item in other objects and it works fine as well 🙂 Small update for anyone having a similar issue - the solution provided by Sam (AggregateTable) requires the "shop_id" column used inside _Fixed_() function to be a character category. In my case it was set as numeric, but all you have to do is change it using the Format operator. Thanks again Sam!

piecze · ‎10-20-2020

Hi everyone, I struggle to create a new calculated item based on the variables shown in the table below. Let's say we have many shops (each matched by single shop_id), offering specific offers (offer_id). These offers are flagged as "1", if they are successful. If at least one of the offers of a shop is successful, then the shop_success variable is flagged as "1" for every row matching the specific shop. Now, I would like to get the sum of distinct values of shop_success for each shop_id. So for this example I would want my calculated item to equal 3 (1 <for shop_id = 1> + 0 <for shop_id = 2> + 1 <for shop_id = 3> + 1 <for shop_id = 4>). What I feel could work is: sum _for all_ (distinct _by group_ (shop_success)), however nested aggregations are not supported, so I suppose that's the challenge I am facing here. I know how to do it with code, but I would like to use the provided table having only these variables. Thanks for any help 🙂

Online Status	Offline
Date Last Visited	‎05-16-2023 11:45 AM

Scoring new data with proc astore and model from MBANALYSIS

Re: Calculation issue (nested aggregation?)

Calculation issue (nested aggregation?)

Re: Calculation issue (nested aggregation?)

Scoring new data with proc astore and model from MBANALYSIS

Scoring new data with proc astore and model from MBANALYSIS

Re: Calculation issue (nested aggregation?)

Calculation issue (nested aggregation?)