Avoiding circular references with calculations on Previous List Item (Numbered List)

chris_liassides

Hello community members! 

 

I am facing an issue which i cannot seem to find a solution to, but that someone might be able to help! This particular exercise started as a way to deal with massive amounts of sparsity in a model used by my employer. The business has a dimension (monthly vintages) representing monthly loan originations. The current approach was to use this as a dimension, alongside Time and a few others, which results in huge amounts of sparsity, as for example, a future loan originated in say Jan 21, still gets provisioned for all items on the time scale the module utilises, even if by definition there is no valid calculation to take place.

I have thought of a way to deal with this using what i would be doing in "excel logic". Basically i figured out a way to calculate the valid intersections of vintages and time, and created a numbered list with the relevant number of intersections. 

 

The problem starts once i try to calculate line items that depend on a previous list member. Anaplan is not able to tell that there is no circularity in the logic (even though the same calculation works just fine when performed in a module with the time dimension). I think this is the same problem discussed here (https://community.anaplan.com/t5/Anaplan-Platform-Discussions/PREVIOUSITEM-PREVIOUS-on-a-list/td-p/24575), but i was unable to apply any of the solutions there (maybe i am just missing something?). 

 

The line items and logic is as follows: 

Good Book - Opening Receivables: Prior List member's closing receivables

New Advances: Linked to input modules

Good Book Payments: Based on an input % of (opening receivables + new advances)

Good book Interest charge: Based on an input % of (opening receivables + new advances)

Flow to Bad Book: Based on an input % of the prior List Member's Good Book's Closing Receivables

Good Book Closing Receivables: Summation of the items above. 

 

I can provide blue print and grid views if required. Would very much appreciate any tips or possible solutions to try out. 

chris_liassides

Hi Bram - many thanks for the quick response! I think this is what one of the suggestions in the thread i linked to was alluding. 

 

I can do Step 1 and 2 (Code property is a number starting from 0 to n possible intersections). I am not sure how to "map" the hierarchy with the time subset, is this done via a line item formatted with time? Or do i need to add the numbered list dimension to the line item (apologies for the basic question here!)? The example would probably help but keen to try it out before asking you to build one! 

 

 

chris_liassides

Hi Bram - thank you again for the reply. I started to do this but realised that it probably wont work for the usecase here.  The user wants the definition of the intersections to be "version" independent (referring to generic list with dummy versions rather than the official "VERSIONS"). 

 

For each version they want to be able to define it as such:

The Line Item "Valid Months & Vintage Instances" calculates the number of intersections based on the inputs. I had created a secondary module which then worked out and allocated to each one of the instances the correct Vintage, month on book, and ultimately calendar month, all of these being version dependent on the original inputs of the user. 

 

 

I then wanted to be able to do the calculations, which forced me to deal with the fact that Anaplan doesn't like using "previous" items in its calculations if the dimension is not "time". 

 

Correct me if i am wrong but if i use the proposed methodology to solve the circular reference here would restrict me to "one" version as that needs to be loaded in the list properties? If i understood it correctly then i would not be able to use this and would need an alternative solution (if one does exist!). 

chris_liassides

Thank you. I will await your message! Thanks for your time thus far - much appreciated!