Skip to content

Mining and Algorithms

September 11, 2012
Amazon

The recommendation engine is the result of some complex algorithms

When we publish articles that look at the performance of T-SQL code, the pieces are very popular. It seems that people are always willing to learn how to write better code, and there is no shortage of people willing to discuss the methods, the results, and the implications of various techniques. It also seems there is no shortage of argument and debate that take place in the discussions as to the best way to solve problems.

However we don’t really understand what will resonate with people, or what things they’d like to learn. What we would like is a system like Amazon’s recommendation engine that might help us improve traffic, and also bring you information that is more relevant and interesting to your career. If you’ve ever purchased something from Amazon and then browsed their site, you might notice the recommendations appearing all over the site. If you’ve received an email from them, you might see some interesting items suggested to you.

Amazon knows their engine is valuable, and it’s constantly being refined. We don’t know exactly what they use to build the recommendations, but they are constantly analyzing the results and tweaking their system. That makes sense for their commerce oriented business, since they can easily measure the results by looking at sales, but it might also make sense for other businesses. Many of us have reports, and other queries used to perform analysis on our data. Do we regularly analyze the performance of those algorithms and look to tweak them?

I’ve found many employers in the past were content to develop a system and leave it alone until some new requirement arose. One company didn’t like that approach, and regularly looked to tweak our systems to try and get more performance out of an inventory analysis application. It was frustrating to regularly rewrite a series of reports, but also challenging to try and find ways to solve a complex problem.

As more and more data professionals begin to work in the analysis and business intelligence areas, learning to build algorithms, and tweak them over time, will be more and more important. It will be challenging, but it should be fun if you are excited by the challenge of improving your algorithms over time.

Steve Jones


The Voice of the DBA Podcasts

We publish three versions of the podcast each day for you to enjoy.

From → Editorial

Comments are closed.

Follow

Get every new post delivered to your Inbox.

Join 4,301 other followers

%d bloggers like this: