Consolidation Weights

I’ve extended the palo mindmap export tool this morning, so that you now have the ability to set element weights within a consolidation. By default the element will have a weight of one, i.e. it will add up within a rollup. However by adding an icon to the mapping you can change the element to have a negative weighting within a rollup.

This should be useful if you are creating a rollups such as:

Gross Profit

Revenue

Cost of Sales

As you are probably aware the accounting formula for the above is

Gross Profit = Revenue – Cost of Sales

Therefore if we wanted to model this using a palo cube we would give the element Cost Of Sales a negative weighting with the weight of minus one. Alternatively this could be done with a rule such as ['Gross Profit'] = Revenue – Cost of Sales. What are the advantages of using a consolidation / rollup ? Well for a start you don’t have to worry about the concept of accelerators. In tools such as palo an accelerator / feeder is used to tell the rules engine that the rule should only operate when a value is populated with data.

Perhaps the best way to explain this concept is to think of something such as Revenue = Price * Volume

In the case of the above calculation you would only want the rules engine to perform this calculation if the volume was > 0 i.e. not null (since anything multiplied by zero is zero) and in a sparse cube this would result in potentially millions of unneccesary calculations. From prior experience accelation is something which people find difficult to understand (and to get correct) so it’s generally best to use a strategy such as weights if possible. If you use rules make sure you accelerate them otherwise you will find the performance of your model to be poor, and for this you only have yourself to blame. If you pull values from other cubes then you really should be thinking out how to tell the engine to deal with this sparsity so it can operate as efficiently as possible.

Shown below is a diagram of how this would be represented in a mindmap.

Whilst I’m at it I thought I might use either the notes or link feature within freemind to allow for the creation of rules within the model.

If anyone has any other ideas about what else could be incorporated then feel free to post a comment. I would particularly be interested to know if anyone has any suggestions for how I could incorporate attributes within the diagram. I feel that the current representation of the cube with it’s hierarchies and levels is pretty effectual and the population of random data allows the user to quickly slice and dice the model and determine what else they would like. The ability to spread data using one of the many front ends is also great. But being able to add attributes would in my opinion be the icing on top of the cake.