Is it worth to require strong data-mining skills from employees?

Question

After I have a small discussion in comments to my recent answer, I decided to ask community:

Is it worth to require strong data-mining skills from employees?

I think that ability to quick find information is one of the most important skills now days.

Would you require employee to demonstrate such skill?

On 99% of programmers interview people ask and answer on questions related to code.

Would you ever ask developer to find solution to a problem using < Name of your favorite search engine >?

EDIT:

Guys, believe me, I know what data-mining is. However I think it is clear what this phrase means in context of my question.

Thanks a lot for your answers!

Answer 1

Most definitely. One of the most important skills in any position related to programming is not "how much information you know", but "how to find information which you don't know". The most obvious way is by using your favourite search engine.

If this was being asked in an interview, I'd try asking them a question about something you're sure they don't know the answer to. Phrase it slightly differently: instead of "What is the answer to XYZ?", make it "How would you find the answer to XYZ?". If their only response is "I don't know anything about XYZ", then that's a bad sign.

Answer 2

I think it would be very worthwhile for developers to be able to quickly find out how to solve problems using resources on the Internet.

A developer who is unable to solve a problem through Internet research can be fairly useless. That's a can be, not an is. As long as the developer is proficient at finding information through other sorts of reference material.

As for "data-mining", I don't think it means what you think it means. Data-mining is extracting relevant information from large sets, such as databases, and is generally used in business, science, and intelligence. There's no need for an employee to be able to do this if they are writing graphic routines, physics engines, etc.

Answer 3

Might I suggest you add "What was the most recent Technical Book you read?" to your interview questioniare.

You'll be either amazed or frigthened to find that 99% of the candidates haven't opened a book in their field since they graduated.

Answer 4

The problem is that most of the techies coming into the field think they can build a software right off of google, the real problem is that few can go beyond that, thus, finding them with iniatives and datamining skills, well, keep looking!

Answer 5

Well as a contributing member to the back discussion, let me clarify my stand on this one. I agree that the ability to sift through massive tomes of information to find something you want is a skill that will take you far... in any field.

That said, We got to live in the real world... I used to be like this sometime back before acceptance of the fact that 'I see bad programmers... everywhere' :) You'll regularly come across 'software engineers' who can't find their way out of a paper house. They outnumber the good ones by a huge factor.

My approach has been to either show them how I do it once.. twice.. 3 strikes and I make a mental note and give up on them. Maybe they'll learn but my experience has been counter to that. Getting angry accomplishes nothing... just ruins my mood/day/team dynamics. Give them the benefit of doubt.. I've myself been stumped sometimes when I dont know the term to search Google for :).. don't penalize people who ask (they're better than the people who come back at the end of the week and say I tried.. but couldn't do it.)

Answer 6

Google is my friend. A great engineer should be able to quickly find out the answer to his problem or at least the best way to move forward on it quickly with only a few tools. Google,... well... and now StackOverflow. :-)

BTW, "Data Mining" is a field all its own. Fraud detection and predictive forecasting are some uses of data mining. In general, the field involves analyzing data, developing a model to describe that data, validating your model against some data you didn't use to train it, then finally using your model to find what you're looking for in future data.

In the specific case of fraud detection, you would create a model of what a fraudulent transaction (or whatever you're detecting) looks like based on "ground truth" or actual fraud cases that have been identified. Then, you would typically use that model to evaluate future transactions and assign a likelihood that they are fraudulent. Then, you hand the ones with high likelihood to real people to investigate. Using data mining techniques to do this can make large companies mega bucks in recovered and/or prevented fraud.

For more information on data mining, KDNuggets is a popular resource. A good overview course to take on data mining is given by Elder Research, Inc. every year. They also do custom training on-site and consulting.

Disclosure: I work for Elder Research, Inc. but I'm a software engineer that does some data mining. Most data miners don't do software engineering; they use data mining software tools.

Answer 7

I usually ask question in interview that i don't expect that somebody know the answer, but i do expect that the answer will be "look in linux/gpl implementation" or something alike.
If you want to check if person have experience in specific field you can ask which resources he knows.
I expect from the person that claims windows kernel experience to know about Mark Russinovich and OSR Online

Answer 8

My two cents:

Data mining is a profession and a lot of computer science theory.

I've written about my opinion on developer's search abilities in my blog:

• Effective developers know how to search the net.

Answer 9

I think there are a couple of different forms of data-mining here:

1) How to get help. Which forums are good ones for getting answers to technical questions? Do you know what kind of support you have from various vendors, which can be useful if you have a problem with one of their products? How much of the local developer community do you turn to if you need help with a problem at work? If you can't get somethign working in a few days what do you do to get past this? It can be a nice question to see if the employee is humble enough to admit that they don't know everything.

2) Statistical calculations. This is slightly different but still can be something that I get asked from time to time. "How many of X do we have on Y?" and so I dig into the data and write various SQL queries on the data to get that answer. It can be fun at times to write these queries and sometimes the answer can be a little scary but this too is a form of data mining. To require strong data-mining here of someone that isn't likely to work with databases may remove lots of qualified candidates. By strong data mining I mean finding means and standard deviations for say sales data over the last 3 years or looking at EBITDA or profits for that matter.