Careers in data science and business intelligence
March 9, 2017 6:40 AM Subscribe
Are you a data analyst/scientist or business intelligence analyst? Do you do data visualization? What specific skills should I develop to switch careers from project management? Are there certifications I should look into getting? What would impress a hiring manager, or at least let me get my foot in the door?
That's pretty much it. I do have some background in GIS, statistics, survey design and analysis, database design and management. I've worked with SPSS, ArcGIS, Access and (obviously) Excel. I've dabbled in Python and (probably irrelevant) Visual Basic but I'm certainly a beginner.
At my last company I tracked and visualized internal business metrics, and I enjoyed that more than anything else I did in project management, so that's one of the reasons why I think I would enjoy BI/data. In grad school (urban planning) we did a lot of GIS/statistical analysis of Census and land use data, and that was much more interesting than e.g. reading about land use law or historic preservation.
Let me know if I'm using the wrong job titles for what I want to do.
That's pretty much it. I do have some background in GIS, statistics, survey design and analysis, database design and management. I've worked with SPSS, ArcGIS, Access and (obviously) Excel. I've dabbled in Python and (probably irrelevant) Visual Basic but I'm certainly a beginner.
At my last company I tracked and visualized internal business metrics, and I enjoyed that more than anything else I did in project management, so that's one of the reasons why I think I would enjoy BI/data. In grad school (urban planning) we did a lot of GIS/statistical analysis of Census and land use data, and that was much more interesting than e.g. reading about land use law or historic preservation.
Let me know if I'm using the wrong job titles for what I want to do.
I work closely with the business intelligence team at my organization and the most important skill for them is writing ability rather than data visualization. This is in the legal industry which has its own unique quirks - so I really could not speak to the career as a whole.
You may want to poke around SCIP. I believe that this is one of the main member organizations for Strategic and Competitive Intelligence Professionals. They regularly have events that would provide for good networking.
posted by rdnnyc at 7:44 AM on March 9, 2017
You may want to poke around SCIP. I believe that this is one of the main member organizations for Strategic and Competitive Intelligence Professionals. They regularly have events that would provide for good networking.
posted by rdnnyc at 7:44 AM on March 9, 2017
Hi AFABulous -- I am not a data scientist, but I work with them a fair bit so I have a little window into that world. My employer is currently building out a more robust data science capability and so is hiring a bunch of data people. For what it's worth, here are a handful of representative data-related openings so you can see what employers are asking for, at least in posted listings. This isn't a plug for my employer (although I am happy here), just wanted to give you some examples that might be helpful. If you want to know more, feel free to memail me.
AVP, Lead Data Scientist
AVP, Big Data Engineer
AVP, Risk Testing Data Analytics & Reporting
posted by monju_bosatsu at 8:21 AM on March 9, 2017
AVP, Lead Data Scientist
AVP, Big Data Engineer
AVP, Risk Testing Data Analytics & Reporting
posted by monju_bosatsu at 8:21 AM on March 9, 2017
I work for a BI/Data science/ Data Viz consulting firm (perhaps the only one that does only focus on that). I also randomly have a masters in Urban Planning too. Me-mail me.
posted by sandmanwv at 8:22 AM on March 9, 2017
posted by sandmanwv at 8:22 AM on March 9, 2017
I consider myself a data scientist and have an interest in visualization as well (although I don't have a ton of opportunities to pursue that interest). I should also note I work in a very SAS heavy environment.
A vast majority of time spent on an analysis project comes down to getting all of the data you need out of whatever system it's stored in and then combining it all together into a usable dataset(s) for whatever analysis you're performing. My main tools here are SAS and SQL. Since SAS is a specialized skill that you'll only really need if you're targeting an employer who uses it, I'd say delve more into building your data manipulation skills in Python and some version of SQL. SAS does offer some free (as in beer) intro products for non-commercial use, so if you are so inclined you can check that out. Personally, I love it, but I also understand that it is its own beast. For the visualization side of things, I'm pretty sure Tableau has a free, public option you might want to explore.
posted by noneuclidean at 8:37 AM on March 9, 2017 [2 favorites]
A vast majority of time spent on an analysis project comes down to getting all of the data you need out of whatever system it's stored in and then combining it all together into a usable dataset(s) for whatever analysis you're performing. My main tools here are SAS and SQL. Since SAS is a specialized skill that you'll only really need if you're targeting an employer who uses it, I'd say delve more into building your data manipulation skills in Python and some version of SQL. SAS does offer some free (as in beer) intro products for non-commercial use, so if you are so inclined you can check that out. Personally, I love it, but I also understand that it is its own beast. For the visualization side of things, I'm pretty sure Tableau has a free, public option you might want to explore.
posted by noneuclidean at 8:37 AM on March 9, 2017 [2 favorites]
My friend went back to university full time and got a 1 year stats master's and now she does exactly this. Not sure where you're located but you could check the universities near by.
posted by St. Peepsburg at 9:27 AM on March 9, 2017
posted by St. Peepsburg at 9:27 AM on March 9, 2017
I also want to note that "Data Scientist" encompasses a wide variety of job responsibilities. For instance, the company I work for doesn't do a lot of machine learning, but we do have a bunch of optimization models (mixed integer programs, mainly) which are fun to work on. I really enjoy building decision support tools but our big analytics department doesn't do much of that because they have other priorities. Luckily I'm in a pretty specialized role in a different department where I get a little more freedom with the projects I work on. Most of our data is also organized in Oracle databases, and we don't do a lot of free text analysis. We're starting to get some better data viz tools, but for a long time it was mainly Excel charts.
My point is just that every company has different needs, and not every data science job is in the tech industry either, so don't get discouraged if you only see one type of data science job; keep looking and you can find a role with facets that interest you.
posted by noneuclidean at 10:32 AM on March 9, 2017
My point is just that every company has different needs, and not every data science job is in the tech industry either, so don't get discouraged if you only see one type of data science job; keep looking and you can find a role with facets that interest you.
posted by noneuclidean at 10:32 AM on March 9, 2017
Best answer: I just landed a job at a city's government development center as a data analyst despite my complete lack of knowledge and experience in urban planning and development. Not QUITE a data scientist YET, but it definitely seems like a junior data scientist role: I will be coordinating a ton of data systems for various department, managing BI Tools, automating reports, running queries with SQL and analyses in SPSS, Python, Excel, etc, creating maps and running spatial analyses in ArcGIS, and developing metrics for impact assessment. I hadn't worked with GIS before this job, which I was honest about in the interviews, but I noted that I desire the role to specifically get hands on training WITH THAT SKILL specifically and was eager and more than willing to learn. My resume is also full of educational experience at both the university level and certifications from places like Data Camp, Coursera, and Udemy, showing that I knew my data/stats stuff and could easily learn more. I also had a portfolio of my visualizations and projects but I never needed to present more than a few examples of my work relative to the position.
What's funny is I have a graduate degree in a non-technical field :p. I am currently enrolled in an online Data Science MS at a decent university and I am sure that also looked impressive or at least like I know what I am doing in terms of data or at least my devotion to the field. I will MOST likely be withdrawing from it or sticking to the graduate certificate after this semester as I am finding it less helpful than it's worth given the price and time. I've honestly learned more from various online courses and books written by professionals in the industry than the homework and projects in these courses, particularly regarding machine learning. I don't exactly recommend these programs yet...they are less than 3 years old, the professors don't really know how to deal with the curriculum or the students, typically they just throw modified problem sets on the course LMS that they've taken from various other universities or websites, T.A.s that are stretched thin manage everything (sometimes), and my experience has been pretty meh. Then again that is just my experience and may have more to do with my learning style or the program or both than anything.
Leverage what you have in terms of your diverse skill set. Data Scientists are expected to be these weird unicorns that know everything from dev programming to data analysis or the idea is that we can learn anything related to informatics/C.S., etc., and no one really knows what they want...but you should be expected to know how to give it to them when they do! You'd be shocked at how baffling executing even a simple pivot table is to the majority of the population. Well..maybe not within HUGE companies like Amazon or what not...but most data science roles people will find themselves in are not going to be that grandiose.
Freelance! Do some projects for random clients, even if the pay is crap at first, to build up experience and a portfolio.
Volunteer! Are there local non-profits that can use some data work? Maybe get in with a company like DataKind for a project or two.
Upload code for projects to GitHub. Credentials are one thing but showcasing actual skill that other data scientists/programmers can evaluate, utilize, and/or build upon in their own work is another.
If you're interested in Machine Learning specifically, sign up and compete in a few Kaggle competitions. I recommend WEKA or Python SciKitLearn although I am basically a novice in both.
Finally make sure you have a convincing story as to why you want to work with data as a career and sell yourself accordingly. Ok the pay is great, but why do you REALLY want to spend 40-60 hours a week for a job that may just be merging spreadsheets and creating line graphs every day (I've literally seen this in a job description filed under "Data Science")?
posted by Young Kullervo at 3:52 PM on March 9, 2017 [2 favorites]
What's funny is I have a graduate degree in a non-technical field :p. I am currently enrolled in an online Data Science MS at a decent university and I am sure that also looked impressive or at least like I know what I am doing in terms of data or at least my devotion to the field. I will MOST likely be withdrawing from it or sticking to the graduate certificate after this semester as I am finding it less helpful than it's worth given the price and time. I've honestly learned more from various online courses and books written by professionals in the industry than the homework and projects in these courses, particularly regarding machine learning. I don't exactly recommend these programs yet...they are less than 3 years old, the professors don't really know how to deal with the curriculum or the students, typically they just throw modified problem sets on the course LMS that they've taken from various other universities or websites, T.A.s that are stretched thin manage everything (sometimes), and my experience has been pretty meh. Then again that is just my experience and may have more to do with my learning style or the program or both than anything.
Leverage what you have in terms of your diverse skill set. Data Scientists are expected to be these weird unicorns that know everything from dev programming to data analysis or the idea is that we can learn anything related to informatics/C.S., etc., and no one really knows what they want...but you should be expected to know how to give it to them when they do! You'd be shocked at how baffling executing even a simple pivot table is to the majority of the population. Well..maybe not within HUGE companies like Amazon or what not...but most data science roles people will find themselves in are not going to be that grandiose.
Freelance! Do some projects for random clients, even if the pay is crap at first, to build up experience and a portfolio.
Volunteer! Are there local non-profits that can use some data work? Maybe get in with a company like DataKind for a project or two.
Upload code for projects to GitHub. Credentials are one thing but showcasing actual skill that other data scientists/programmers can evaluate, utilize, and/or build upon in their own work is another.
If you're interested in Machine Learning specifically, sign up and compete in a few Kaggle competitions. I recommend WEKA or Python SciKitLearn although I am basically a novice in both.
Finally make sure you have a convincing story as to why you want to work with data as a career and sell yourself accordingly. Ok the pay is great, but why do you REALLY want to spend 40-60 hours a week for a job that may just be merging spreadsheets and creating line graphs every day (I've literally seen this in a job description filed under "Data Science")?
posted by Young Kullervo at 3:52 PM on March 9, 2017 [2 favorites]
Am a data scientist. You seem well positioned for BI analyst roles. Learn how to work with databases (suggest PostgreSQL) and get really really good at SQL; develop software carptentry skills and develop your program ability in Python; try to get exposure to some BI tools like Tableau and Pentaho. Good luck!
posted by PercussivePaul at 4:14 PM on March 9, 2017 [1 favorite]
posted by PercussivePaul at 4:14 PM on March 9, 2017 [1 favorite]
I am not a data scientist but I am around a number of people who use Python for data science. If the language interests you, it's apparently used for a lot of data analysis. The kind of programming data people do isn't very complicated (from what I've seen they hardly use classes, for instance) so you are probably set in terms of your skills there. They are very fond of Jupyter notebooks for development instead of a text editor or IDE.
The most commonly used libraries seem to by numpy and pandas (which produces spreadsheet-like dataframes). For visualizations, many people seem to use matplotlib and seaborn. You might also want to play with plot.ly, folium (for maps), and bokeh.
posted by Lycaste at 2:43 PM on March 10, 2017 [2 favorites]
The most commonly used libraries seem to by numpy and pandas (which produces spreadsheet-like dataframes). For visualizations, many people seem to use matplotlib and seaborn. You might also want to play with plot.ly, folium (for maps), and bokeh.
posted by Lycaste at 2:43 PM on March 10, 2017 [2 favorites]
« Older Theology and history of fasting | What side dish to bring for a camping dinner feast... Newer »
This thread is closed to new comments.
What it's like to be on the data science job market
Data Scientists and Their Work
posted by rada at 7:35 AM on March 9, 2017 [1 favorite]