|
Jon Joe posted:I have a Master's degree tailor made for doing data science but because it's not named data science but instead psychology nobody ever believes me and I have literally never gotten a data science interview in *checks* over a year despite having used Python, R, and SPSS to comfortable levels. SPSS is used a lot in academia/market research but not much outside of it. It depends how you're using it - crosstabs, modeling, cleaning, etc... - try to write that more broadly and platform agnostic. Just assume that recruiters and resume parsers are only there to narrow down the pool, so you need to write a resume to get past them and get read by a real hiring manager, not necessarily the best representation of your skillset. Vomik posted:All fair points and of course the whole association with "less than savory" aspects of quantitative finance. If you search for something like "SQL" on a job site, you'll find large companies in any industry you can imagine. Anything retail/ecommerce will have huge supply chain and digital analytics operations. Who's sitting on so much customer data that they're regulated to hell and back? Finance, telecom, and health care. Between those four areas, you have a really big chunk of the Fortune 500. 6. McKesson
|
#
¿
Dec 29, 2018 07:07
|
|
|
|
| # ¿ May 22, 2024 13:26 |
|
|
SpaceSDoorGunner posted:I'm currently active duty military but I spend most of my time at work on call, so I've started studying finance, which then lead me to learning about simple forecasting methods, which lead me into statistics, where I then realized that the only way anyone makes any forecasting stuff work enough to be not totally worthless is through methods that require massive amounts of computing power, which has taken me down the road of learning python and "datascience" though I don't know if it's cool to say "datascience" anymore. It depends on the area, but I don't really agree with this. A lot of models are just doing regression on data sets on a few thousand records, which can be done locally on your laptop. SAS can handle more but is probably mostly a relic at this point. So that means realistically, you're using something like R or Python sitting on top of a database, or directly working with the data in AWS, Google, or Azure. And what's great about the latter options is you can buy them pretty cheaply, especially if you're willing to use a spot instance of EC2. I know companies who can get dirt cheap computing hours if they can manage the EC2 auction system, and even vanilla AWS is a quantum leap better/cheaper from what was common place just a few years ago. ultrafilter posted:It probably will be. I don't have the exact numbers, but from the last BurtchWorks report I think something like 45% of data scientists have a PhD, 50% have a master's but no PhD, 5% have only a bachelor's, and the people who don't have any degree are rounding error. I would expect the percentage of people without a PhD will go up over time, but most of those positions will go to master's holders. Agree on not liking the masters programs generally. I don't see why those couldn't be offered as vocational training. They're fine-ish, but we need to break this mentality of degree inflation - those programs are mainly to reassure mid-career transitioners from other fields, I don't think they actually need them in terms of skills - maybe for networking, or forcing them to actually do the work. Those numbers you quoted seem WAY too high though, although it definitely depends how you define the occupation. I know a lot of people, including myself, who have just a BA and do just fine. (I kind of went academia -> private sector.)
|
|
#
¿
Dec 29, 2018 07:18
|
|
|
Suspicious Lump posted:So probably a silly question but apart from extracting data from a server, is SQL used for anything else? I'm trying to understand why SQL skills are so sort after. It's used for building a lot of reporting in this context, others have mentioned the database administration or software development side of things. From what I've seen the majority of data scientists don't have a background in stats, but will know how to build predictive models through whatever technique in some of those languages.
|
|
#
¿
Jan 14, 2019 04:02
|
|
|
Honestly a lot of this sounds really shady. Half the people I know in the field are self taught and have made it largely on their own personal momentum. If you're really apprehensive and think you need a kick start if only for networking, fine, but I don't want people to be discouraged and think they HAVE to do this or else.
|
|
#
¿
Feb 17, 2019 05:26
|
|
|
MickeyFinn posted:When did they get started in the field and what were they doing before? What is "personal momentum?" Running the gamut for the past decade. Personal momentum is you don't sit around and do exactly what you're told and nothing else. Data scientists need to be genuinely curious, and the type of people are just going to teach themselves something new, like a language or a package, out of a blue. The former yeah, but I don't know anyone who went through a boot camp. I'm in corporate America (past 5 years, startup before), and I don't know how it's different in tech or at smaller companies. ultrafilter posted:Anyone looking to break into data science right now should read Data science is different now. It's not a happy read, but it's full of things you need to hear. That's a weird quote, as Hive is widely used. It's true that there's an oversupply at one specific point where people have been susceptible to get rich quick schemes, but there is massive undersupply elsewhere *at the entry level*, and massive undersupply at the higher levels. There are 100 people applying to one opening because in many cases, they aren't qualified. And the piece is spot on that a ton of your team is spending doing data cleaning type things which honestly, a lot of the modeling/machine learning types are often bad at. But that doesn't have a whole lot to do with a CS degree. She has a very skewed viewpoint towards tech which is not representative of the entire field. Her advice at the end is mainly good. zmcnulty posted:So I have couple questions: Document things, including properly staffing on metadata Hire data engineers who know how to do things like ETL I doubt either will be done, or done adequately. Learning SQL is easy, learning the others can definitely be done quickly but it depends on your level of commitment. If you're just moonlighting, it's hard for them to sink in. Kim Jong Il fucked around with this message at 00:26 on Apr 22, 2019 |
|
#
¿
Apr 22, 2019 00:22
|
|