Can someone please explain what pic related means in day to day terms
If it's your job, wtf do you actually even do for 40 hours a week
If it's your job your job is maintaining the computers that do it.
Data mining is something computers do and if you 'data mine' you're maintaining the computers.
I have a bit of experience with web parsers, this is what I imagine it would look like:
check error reports on failed data minings
- parser errors
- network errors
- analytics errors
fix those errors, improve parsers, implement new parsers, research new format standards
there goes your 40 hours
>>56660777
>failed data minings
What is a "data mining" and what makes it a success or failure? Or do you mean systems failure?
What actual business cases do these people solve?
>>56661008
You have lots of data in a format you can't use very well (imagine tweets that contain some hashtag you are interested in), then you research and implement software to extract usable information from this data. This information will then go to business analysts to help them make decisions
>>56660727
"Data mining" is an attempt to apply Shinola to the turd that was called "Database admin", specifically the report writing part of a database admin's job.
Oops, did I hit a nerve?
Prove me wrong, then.
>>56663662
Wrong
"Data mining" is [buzzwords]transforming data into information[/buzzwords].
There are multiple ways to do this, some of then which work, others that don't. Not all datasets are appropriate for this.
Also, a database admin's job is entirely different than that of a [buzzwords]data scientist / data miner[/buzzwords]
>>56660727
I've only done an internship and a masters, but here's my best guess:
- Step 1: Try feeding data into some classifier or regression model
- Step 2: Battle with getting the data flow to work
- Step 3: Make sure the analysis is actually working
- Step 4: Produce technical report and analytical report
- Step 5: Improve model
- Step 6: Repeat Steps 2-5 until you have new data to work with, system changes (and needs fixing), or you get a result you can't improve.
>>56663807
Gosh, you used the buzzword field. That sure proved me wrong.
If some business wants to do some "data mining", are they going to hire a "data miner"? Or are they just going to pile that shit onto their database admin(s)?
>>56664149
if they just give the problem to the database admin, there is little chance that he will be motivated, or even competent to do the job. Failure is guaranteed.
You want to hire / contract a specialist / a team of specialists
Data mining usually deals with finding hidden pattern.
Take example of your credit card purchases. You usually purchase from target but all electronics strictly from Walmart.
You also pay your bills on time, but when you purchase electronics you miss it.
credit card company can analyze the pattern by looking at your purchase history. They can sell your info to loan companies saying if you target this guy around certain time, he's most likely to take out loan.
May be it can promote the Walmart electronic sell so you are more likely to buy more.
If you notice it's not just regular analysis like which customer is biggest spender or such. It's pretty specific pattern identified which otherwise would not be visible.