HN Explorer

Ask HN: How to find business problems to solve as a data scientist?

19 points by stevofolife ↗ HN
To provide some context: Our entire data scientist team is about to join our parent company. In anticipation, how can we determine a business problem to solve given that 1) we don't want to do the dirty work for other data science teams or steal their work, and 2) there is no clear mandates being passed down to us from the higher up.

Obvious processes to go through include talking to all the data science team and learn more about their work and future pipeline, speaking with the business folks if possible to understand their needs. But is there something else that we can do that hasn't been listed yet?

16 comments

[ 2.0 ms ] story [ 41.6 ms ] thread
ilikeatari
#2 is somewhat problematic. I would suggest finding someone in your organization that truly knows the business you are in and it’s day to day challenges. Identifying this person or people might be hard but it is crucial. Establishing true and open relationship with this person or group is the next step. It has to be genuine. Once that’s established you will see the stream of real pain points and it will be your job to map them into data science opportunities. An alternative option is to shadow day to day ops crew and distill requirements that way.
probinso
This is kind of significantly dependent on the type of work you're doing.

When in doubt, nearly every company has bottlenecks we're either a human or a series of humans have to perform parts of a single task. Usually a data science team can find ways to reduce the number of hands in cookie jar.

If the data science team has any natural language processing expertise, then usually there is some work in organizing text documents of some sort.

Data scientists are kind of a general title oh, so you're going to want to know what specializations they might be coming in with. Many data scientists got their starts in business intelligence. Those data scientists will probably have strong opinions on how they can inform the business.

espeed
Watch the news and solve problems they can't seem to solve.
iamnothere123
LOL, you know you are doing a BS job when you have to look for a problem to solve
dang
Please don't cross into personal attack on HN.
iamnothere123
Its not a personal attack, Its just true
dang
True things can be personal attacks.
iamnothere123
Describing the state of something is not an attack
dang
You can't reduce communication to simple rules like that. It depends on who is communicating what to whom and why.

For example, when a middle school kid has acne and another kid goes on about their face, that is not a neutral "describing the state of something". Everything they say may be perfectly true, yet it is certainly a personal attack—and the more cruel, the more true it is.

iamnothere123
Seriously are you taking child behavior as an example?

"Argument by analogy" look it up

dang
Adults are no different from children in this respect, and the example is particularly easy to relate to.

Please don't post rude swipes in HN comments. It's against the site guidelines (https://news.ycombinator.com/newsguidelines.html), even when addressing a moderator.

iamnothere123
The bar for what is offensive is very low it seems. Any good discussion will always end up offending some one. Just don't take it too personally
dang
I wouldn't personally frame this in terms of "the bar for what is offensive", with its overtones of "what you can't say" and all of that. It's rather that we want users to remain kind to each other, when talking about difficult things. The intention is not to exclude or suppress communication, but to enhance it.

Still, you're welcome to frame it in whatever way works for you. The point is that you need to follow the site guidelines if you want to continue posting here. If you'd please review them and take their spirit more to heart, we'd be grateful.

natalyarostova
Try to see the world through the eyes of the product managers/finance/ops guys (pick whichever makes the most sense, or all of them). Then pour through their excel workbooks to look for something that they are currently approximating, but which would benefit from a rigorous approach.