Hi Kshap, The objective function is to minimize the cost of inventory. The constraints are: Achieve a set level of contribution Include every SKU in your analysis. In the classic Knapsack problem every fruit has a weight and a caloric value. Similarly in my problem, SKU has a cost and a value contribution to service level. The tricky difference is that the value of contribution is not scalar. It's based on Poisson distribution. So 1 unit could contribute 30% probability of service level, while 2 units together would have combined contribution 45%, or more accurately not 60%. (I translate probability into contribution value) Therefore as the procedure goes through different scenarios of potential solutions, N_unit contribution and 1unit contribution cannot occur for the same SKU in the same scenario. My problem is actually very similar to https://communities.sas.com/t5/SAS-IML-Software-and-Matrix/Variation-of-Knapsack-Problem/td-p/569457. In that example there are 3 arrays (he calls them lists). To do the same thing I would need at least 6000 arrays for each SKU. I hope that clarifies it better. @Ksharp @Rick_SAS Rick, I am afraid I am the last of the Mohicans here, the younger generation shy away from SAS in favor of Python and R. Yesterday I asked around for advice in other groups, but those users are just running old scripts that were not developed by them.
... View more
