Optimising resources. How ?

Replies to This Discussion

Permalink Reply by Huayin Wang on July 23, 2008 at 10:18pm

What type of resource allocation is it? In my current work, I do optimal budget allocation. The set up of the problem is roughly this: suppose you have a total of $x of budget, and further more, suppose you have several projects where you can allocation you budget to. Each project will provide you some returns. What's the optimal way to allocation your budget?

If you total return on investment is the sum of return from all your projects and return on each project is a function of the budget allocated to it. This is really a simple optimization setup. The solution is: the marginal return on investment should be equal for all projects.

This should be pretty easy to carry it out, except in really, you do not know the ROI function for each projects. That's the tricky part.

▶ Reply to This

Permalink Reply by Yves Zieba on July 26, 2008 at 11:45pm

hi Huayin Wang,

Thanks for your answers and for the help. The setup you describe is very similar to the way we operate today. You are right, the problem of resource allocation is in the framework of budget allocation. To give you some context, we have multiple projects (we call them surges), which are all decided by the company 's strategy.
We are well equipped to analyse the return on investment by project or by product.
What we are less equipped with, is to have a way to calculate return on resource investment. So if the company wants to know if an investment in let's say 3 more sales people in a particular territory, the company has no direct way to know if this investment was worth it or not. So when it comes to design team or individual objectives, it is quite difficult to know how many projects you can allocate to them and to measure their team and individual productivity.
So basically the key decision that my piece of work is supporting, is how many people in each key role should we have to successfully deliver on the company objectives.

▶ Reply to This

Permalink Reply by Sean Salleh on September 29, 2009 at 8:09pm

Hello Yves.

The problem you describe here, in short I would assume, is rather common among many businesses, as anyone reading this may know. I don't know your background; so please excuse the following if it is rather elementary.

Without knowing much else about the problem--the task of attributing individual and team performance and designing teams--it should be clear that we can approach the problem from a predictive then optimization perspective. So, we can try to build regression models to try to assess how much sales revenue may be attributed to a specific salesperson S in territory T.

Okay, suppose we have a good regression model such that if we know certain characteristics of a salesperson and the territory they may be assigned to, then we can estimate the amount of revenue that may be attributed to their performance in that territory (or any territory, hopefully). Say we can do this for each salesperson.

Next, we can formulate an optimization model in which our goal is to determine how best to assign certain salespeople to all territories so that the aggregate ROI (or RORI = return on resource investment) is maximized, given budget and other constraints.

From here, we should get decent results, from which management should then use to help make good decisions. Some people may say that the predictive portion of this solution approach is the most difficult, and I agree for the most part. However, care and creativity must be used in formulating the optimization model so that good optimization algorithms may solve the problem optimally or near-optimal.

Of course, there are other solution approaches to this problem that others may suggest, but a prediction then optimization approach is common.

▶ Reply to This