Hail to the champions, the University of Kentucky. I only got one of the four Final Four picks correct, but luckily that turned out to be the champion. My SAS-generated picks turned out to be in the 71st percentile of all those who participated in the Yahoo NCAA challenge. Not bad.

Careful readers may have noticed that my code had a somewhat serious bug in it: the exponent in the “pythagorean” formula that I relied upon was supposed to be 11.5, but I had 1.5 in my code. Oh well!

### Like this:

Like Loading...

*Related*

Hey Nate,

I didn’t know how can i address you, so i posted my question here.

Recently i have started using Microsoft solver, and i have noticed your blog, which have helped me a lot.

While trying to model my problem i had some problems, and was thinking maybe you can throw light on this matter.

My domain problem is as follows:

I want to assemble computer from different components(CPU, Motherboard, Memory ,Hard drive, etc… )

1. Each component can be bought from different vendors(HP,ASUS,INTEL, etc..)

2. Each vendor has some or all the components(I need to minimize the number of different vendors).

3. I want to find which Vendor to buy from for each component(an IntegerRange(0,1) decision)

4. Each vendor is also a decision variable- IntegerRange(0,1)

5. Each Vendor has Rating that ranges between 1-5.

6. Each assembled computer has rank according to the following:

40% is 1 divided by the Total Price.

30% is 1 divided by the number of different Vendors

30% is the average Vendors rating of the assembled computer

Remark: The reason i divide 1 by Total Price is that the rank is in inverse proportion to the Total Price and also to the number of different Vendors

For example: i choose to buy CPU from INTEL- costs 100$ , Motherboard from

ASUS- costs 70$.

. INTEL is rated 5, and ASUS is 4.

TotalPrice is 170$, NumberOfDifferentVendors = 2, VendorRating =

(4+5)/2 = 4.5

Rank = 0.4 * (1/TotalPrice) + 0.3 * (1/NumberOfDifferentVendors) +

0.3 * VendorsRating

Rank = 0.4 * (1/170) + 0.3 * (1/2) + 0.3 * 4.5 = 1.502353

7. I have issue with normalizing the rank. the price is not on the same scale as rating, (maybe you have suggestions)

8. I need to find several best ranked assembled computers according to formula described above.

i.e- for example: find the 5 most minimized rank assembled computers .

9. I have manged to find a minimized Computer, but only by the price not by the rank formula,

10. I couldn’t also find a technique to produce several minimized solutions.

I have consulted with other people in the field, that unfortunately, were unable to help me.

Is there a reasonable model for this, that you know of, which can be solved with MFS or other products?

I highly appreciate your answer.

Sincerely,

Paul Resnik.

Any idea???