Optimizing Models, blueprint Analysis
Hi all, I am beginning my journey as a model builder in a company where one of the first task is to try to optimize the current Anaplan Models,
I've followed the instructions given here to try to reduce the cell count in a specific module:
- Formula repeated several times in different line items and modules. And one those line items has more that +3M count cell.
CODE(ITEM('SKU CODE P5'))
- The formula only reference a list called "SKU Code P5"
So I took one of my archived models, changed the mode to standard and then created a module dimensioned only by that list and then replacing the formula by a reference to my new module.
I was expecting to reduce the cell count as explained in the example section of the link shared, but I am getting the same +3M cell count.
The following image show the reference of my new line item.
The first line item of the following image has the formula and the last one has the reference, and both of them have the same cell count.
Not sure if I am doing something wrong.
Any advise would be highly appreciated.
Lets understand that, Cell count is based on Number of dimensions and Summary(if you turn of the summary it acts the list as virtually flat list, thus reducing the hierarchy count).
Lets say Module A has 3 dimensions.
Lets assume each dimension has 2 items. so the total count is 2*2*2 = 8.
In the article that you referred says that if there are several line items with same formula, it will increase the redundancy and thus increasing the cell count.
Ex: Line item A - Code(item(list)) - has 10 cells, if we have similar line items with same formula, it is just repetition.
With the blueprint analysis we can find how many such similar line items with same formula are there and we can delete everything except one that we will use for reference.
while in development generally people cannot check how many line items they build with same formula, during the end phase of development we usually do this kind of optimization check.
referencing and writing the formula with same dimensions have same cells count.(if we have same dimensions)
The purpose of blueprint analysis is to remove such repeated line items and refer from only one line item.
The reason why the SKU ID is still 3M cell count is that you still have Time associated with that line item (as @ManjunathKN ) spoke about. When attempting to optimize a model, you might want to give this a read as well:
Thank you @ManjunathKN so as long as I have the same dimensions in my module it will have the same cell count, so I guess I need a different approach if I want to reduce the current size of the model.0
Thank you @rob_marshall I will review the link as I need to find ways of reducing the size model.0
The number of line items will also affect cell count. For example if you manage to halve the number of line items you'll half the cell count.1
The key here is thinking about what is relevant to the outcome of the function and to only include dimensions that meet this criteria.
Your line of thinking is correct in that this line item is only relevant to SK Code P5 and should therefore only be dimensioned by this list.
It is not clear from the screenshot what is included in this line item but if you open the blueprint and click on the ellipsis on the 'Applies to' section you will see that you can deselect all other lists to only have SKU Code P5 select.
Next check that time is set to 'Not applicable' and there are no versions assigned to the line item.
However, I would advise that you create a separate module only dimensioned by SKU Code P5 and create the code reference there and direct your formula in the target module to this new module directly. You may need to reference this code in other parts of your model and we should aim to create once in a single location and reference it directly from multiple targets. This will ensure your model is efficient.
You should create these 'properties' modules for all lists where you need properties such as code and mappings.
Remember that we can reference other modules that share dimensionality to retrieve properties such as code without having to create the reference directly in the target. This is often a common error for new model builders trying to recreate Excel models in Anaplan.0
Yes, There are several ways of doing it for several sparse outcomes.
https://learning.anaplan.com/course/view.php?id=519 - Check this article for reducing the sparse cells and Yes, Please also refer the one @rob_marshall had mentioned about model optimization.