Computer science/ maths/stats - honest advice required on undergrad degree vs. conversion master's?

[pomBearPooPouffe]

Computery types, or academic types generally - please help!

I am switching fields to IT/CS because I can't get a job as an academic biologist, but there are plenty of jobs around in IT and CS/maths/stats type research & development. I have been planning to do an undergrad degree majoring in computer science, with maths and stats subjects.

However, all the CS/IT/maths people here reckon that with a previous career* as an academic biologist, I should just do a one-year master's by thesis and attend a few high-level undergrad courses in maths and computer science. They all say undergrad is unnecessary.

I think this is ridiculous, because I have no useful background whatsoever in maths, stats or CS. I honestly have no idea how any of it fits together. While CS isn't maths (except in a few areas) it assumes basic competence with things like calculus, geometry, matrix algebra etc. - most of which I've never done and have absolutely no clue about.

I did badly in the "maths for idiots" option at school (about equivalent to a B at A-level), I flunked first year maths at university (56 % in the "maths for biologists" stream, probably because I went to almost no lectures and did no assignments); and I've never done any statistics at all beyond the complete basics taught to biologists. As for CS, I've done a couple of very basic introductory programming courses (Python, BASH) and really enjoyed them. I've always been the most analytically-minded person who "gets" how to deal with data, in every biology lab I've worked in (which of course isn't saying much when you think about the analytical skills of most biologists). However, being able to understand how analyses fit together, use command-line software, and follow the first 3 chapters of a Python textbook isn't "having a background in programming", it's parroting syntax without understanding a word of what's going on.

I think that if I followed the advice of doing a masters, I wouldn't understand a word of what they were talking about in a 3rd year undergrad class in maths or CS - I can't even really remember how to do basic high school stuff like quadratic equations or calculus. I do think I'd survive undergrad maths and stats because I can pick stuff up quickly when I need to, but I do think I'd need to do the full undergrad degree to have the faintest clue what's going on.

Am i mad, or are the people offering advice just not thinking through precisely how much maths etc. they did themselves at high school level? My DH for example did university maths while at high school, as did many of his friends and colleagues - he genuinely doesn't seem to understand that there are fully-functioning academics out there who don't know how to solve a quadratic equation...

*Previous career = working in academia as a biologist since 1998, PhD awarded in 2004, postdocs for 7 years, got to lecturer level with lots of experience, but couldn't get made permanent because of lack of jobs available here, went back to postdoc-ing, hated it, gave up. Could probably get a job as a biologist if I were in the UK, USA, Europe etc., but not where I currently am, as there are too many biologists here and few jobs generally.

From the sounds of it, you're not really looking for an indept course on actually computer science, as that would be the more maths and algorithmic side of things.
If you want to learn how to code, there are definitely other more useful options, such as MOOCs, websites and bootcamps which may give you a firmer grasp on functional programming without the underlying principles and theory that a CS undergraduate would focus on.
I agree with your friends to an extent and think a Master's/Research project would probably be the way to go.
Hope this helps.

Thanks for the perspective Missinvisible7 - are you coming at this form the perspective of someone who works in IT or as the analyst/programmer for a larger research group in a university, or...?

I'm asking because I'm trying to get a feel for where actual CS, rather than knowing a couple of languages, is important.

I kind of assume that if I'm going to be the analyst person in a large medical research group, for example, then I need to know how to do scripting to handle all their sequence data, but I'll also need to know how to write code extremely efficiently, how to implement complex maths and stats, how to parallellize analyses, potentially how to cope with data security issues over networks, etc - and all of that is CS rather than IT. (Just FYI i'm in a small university town that has a huge medical school that's very strong on research, particularly data-driven stuff - so this is a realistic prospect for where I might get a job in a few years' time)

Obviously if I end up being the IT person in some government office then I'll spend most of the day installing Windows and antivirus updates and switching computers on and off again - but I'm hoping that if I end up doing IT then I will at least do it in a context where keeping abreast of stuff like cloud computing might come in handy...

I'm currently a CS undergraduate myself and my course is pretty theoretical. If you do think that those skills may be most useful in the sort of analysis work you want to go into, then yes maybe an undergraduate may be the best way to go. And although a CS degree won't teach you a range of languages, you will have the skills to pick up scripting quite easily. I just wanted to point out that it may take longer and be more expensive than some of the afformentioned methods.

Thanks for that. I think the undergrad course where I am is somewhat more practical than some of the extremely theoretical courses on offer in the UK - which is why I'd see it as a necessity to add in maths and stats as minor subjects, because it's not necessarily the case that the CS grads from here would have appropriate skills in maths and stats to become data analysts.

I definitely agree that a 1-year master's is less expensive than a 3 or 4-year undergrad, and that supplementing with MOOCS, online bootcamps, etc. is definitely useful. But, a 1-year masters' would be just OK for turning me into a crap bioinformatician, and would be pretty much useless for anything else.

I actually think the extra expense of undergrad would be worth it in this case.

If you want to be a statistician, wouldn't a maths/stats degree with comp sci on the side be more appropriate?

I used to work for a CRO and the database/programming/stats software was SAS, which you wouldn't learn on a comp sci degree and which would make an in depth knowledge of Java superfluous.

Maybe contact the medical company and ask them what they are looking for?

I'd go for a full degree.

I have a BSc in computer science and what I learnt can not be learnt via MOOCs etc.

It was a really good, really interesting degree which has been very useful and opened all the career doors I wanted it to.

For a computer programming job interview you will generally be asked to code and you will also be asked more theoretical questions.

Plus programming is a craft. It takes years to get good at it. 1 year from a standing start is just not long enough.

Noblegiraffe - in biology, because it's now based so heavily on sequence data and it's so easy to generate WAY TOO MUCH sequence data, there are now lots of jobs for "analyst who can write programs to implement the analyses, and also knows what maths and stats should be being used for this kind of data" - the old "biostatistician" position in public health departments has broadened somewhat both in the skills required and the biological applications of the analyst's work.

I'd assume a CS degree would give me the ability to know how to pick up languages, and that the appropriateness of a language would depend on the application. That's why I want a generalist theoretical framework that allows me to choose in an informed way between FORTRAN, Prolog, C, Java, Ruby, Python, Swift, whatever, and pick the syntax up as I go along; rather than for example learning Python and Java without the theoretical framework, and subsequently trying to hammer at whatever comes up in future with the two languages I know, whether or not they're appropriate.

The medical school here would probably want to employ analytical people who can move on at research pace - so they wouldn't be looking for perfect knowledge of a given language, they'd be looking for your ability (across many different projects) to justify your choice of experimental design, maths, stats, and implementation - and then your ability to go away and get it done, possibly also by hiring an underling to do the coding who's a specialist in the language(s) you've chosen.

Wildstallions I agree.

I would be far more concerned about getting a qualification from the right place. £27000 for a degree which is of very low value seems like a huge waste.

What kind of jobs do you want at the end? Who would you like to work for?

If you can answer those then go and talk to them, and get some careers advice on which courses they are most likely to recruit from. What they think of your plans, and why they like their preferred places/courses.

There are specialist conversion courses out there, and they might meet your needs more closely.

I have a friend who did a 1 year Masters conversion, having a first and Masters degrees in biology. He became a highly sought after Computer Professional, not just working in bio-informatics. But it was a proper conversion course.

I would think of something like Birmingham's www.birmingham.ac.uk/postgraduate/courses/taught/computer-science/computer-science.aspx
Or Oxford's part time www.cs.ox.ac.uk/softeng/

Or for the specialist area you are seeking, Birbeck's mscb.cryst.bbk.ac.uk/

thanks mummytime - unfortunately I'm now not in the UK - and live in the best university town in my country - so choice of uni isn't really an issue, the local one is the one to go for, thankfully. That said, I wouldn't mind going back to Oxford!

I have a computer science degree and worked in the software industry for the last 35 years. My degree helped very little, except to get my first job. I've worked in big teams and usually there are only about 10% who are computer science graduates. On the occasions someone needed advice on a particular technique I was available to ask. The air traffic control software I worked on had 2 computer science graduates, 2 mathematics graduates, 1 psychology graduate, 3 electrical engineering, 2 controllers, 1 chemist and 1 farmer (he came in on day to see if he should change careers, still there 20 years on).

The point is - learn programming, then the other stuff like standards and testing and version control can be learnt on the job. Learn programming by reading or going on a programming course then reading code.

Most of the time in a software job you won't be coding. You will be reading/writing documents, planning, testing, debugging, consulting with users and thinking of ways to kill the project manager without getting caught.

Look at the skills you have and check those against the required skills that will tell you what you need to work on.

A taught masters might be a good idea. Not a research masters - you've ibviously not got the background for that.

What about bioinformatics?

I think you may find an undergrad degree too easy. I know I did when I did some OU modules in a different field. My learning skills were good so the course was way too slow for me.

Look at taught masters near you so you can get in-person support if required.

I have no expertise but last month I was speaking to a medical statistician, working on clinical trials, who had an undergraduate degree in marine biology.
My son is a mathematician who has no idea how medical trials work, no interest and no expertise, so I also wonder if a taught masters is what you need.
Again, I know nothing.

This sort of thing? www.ed.ac.uk/biology/postgraduate/taught-programmes/bioinformatics/programme-structure/taught-courses

7to25 - thanks for the encouragement
I have relevant background in biology to work in the biggest likely employer here, which is the medical school and other similar university departments. i'd like to pick up the maths/stats/CS required to work there or somewhere similar (for example, the local equivalent of NICE).

Museummum - taught masters' would be a good idea, but there aren't any here. I agree aspects of undergrad CS would be easy, but that gives me time to concentrate on the stuff I'm likely to find hard, e.g. maths.

cdtaylornats thanks- that is very useful. This is a bit of a silly question - but do you feel that you'd be equally happy in a research environment as in a software company type environment? Given that I am in a small town (and I don't want to move at the moment, for various reasons) I'm quite worried that learning to program in a few languages won't be enough to get me anywhere in the job market. Thus I'm trying to find the most efficient route to being employable at a basic level in whatever comes up, whether that's as a random codemonkey doing software kind of stuff, or as a research associate handling the med school's data. Given that I'm 18 years older than most new graduates, I need to be better than them in areas where my background doesn't count, to get a job. I had assumed the obvious route to being better would be doing well in an undergrad degree. But you seem to agree with peoplein RL who have offered advice. Hmm.

I'm a software developer, with a degree in Electrical Engineering, and a PhD in artificial intelligence. There are many many computer science graduate, and if you google, you'll know that it's one of the least employable undergraduate courses. I reckon the problem is that a lot of graduates can't actually program. On the other hand, if you can land your first job as a developer, you won't need to have a CS degree. Every interview I have been in requires a programming test. Some of them very involved. And you would need an understanding of algorithms and data structures background to pass them. (It usually involves solving problems that needs knowledge of data structures).

I would not recommend you getting a job as a developer. You have a PhD in biology and you should definitely make use of that advantage. I think you have identify your niche as an analysts for biology. As you say, there are now a lot of data to be analysed in biology. And this dependence will only increase. You will simply need to be able to script/code well enough to process biological data. Have you talked to any analysts in your field? Can you ask them what qualifications they have? Do they have any recommendation on how to get into the field?

As a engineer and a software developer, I reckon I'll need to learn the 'language' of biology before I'll be able to work in a research environment analysing your data. We don't just blindly implement what's asked of us. We recommend solutions, as a lot of time, the person making the request for a feature have no idea what's possible. You already have the advantage in understanding the field. Do not give that up!

Basically my worry is this if you go through an undergraduate CS course.

I used to work for a CRO and the database/programming/stats software was SAS, which you wouldn't learn on a comp sci degree and which would make an in depth knowledge of Java superfluous.

As a graduate of CS, you should be able to pick up languages through self learning. I've have never been taught a programming language. They are actually all very easy. All you need is the online tutorial for that language. I would say, as someone working in the area, all those courses you have to pay to learn a language is a rip off. You learn the theoretical stuff in universities which no website teaches you.

Thanks Headdreamer. Do you agree with what I said upthread about use of languages

I'd assume a CS degree would give me the ability to know how to pick up languages, and that the appropriateness of a language would depend on the application. That's why I want a generalist theoretical framework that allows me to choose in an informed way between FORTRAN, Prolog, C, Java, Ruby, Python, Swift, whatever, and pick the syntax up as I go along; rather than for example learning Python and Java without the theoretical framework, and subsequently trying to hammer at whatever comes up in future with the two languages I know, whether or not they're appropriate.

...or am I assigning way too much importance to the necessity of choosing a language for effective/efficient programming? What about data structures, algorithms - should I be picking those up from MOOCs instead? Presumably I'd need some knowledge of those, of networks, etc.

If I am going to stick with the idea of having a data analyst niche, I probably need a degree majoring in stats with maths and CS on the side. But trying to stay open to a wider job market, that's less saleable than a CS degree here (I agree about employability stats in the UK for CS degrees; it is quite different here because there is far less population, the economy is far more stagnant, and it's far more socially stratified, so there are almost no non-graduates who are looking for jobs in areas requiring specialist knowledge - so actually CS degrees have the best employment stats here of pretty much any degree).

Have you looked into bioinformatics courses. I think some maybe distance learning if your own university doesn't offer them.

When I worked in this area the group was roughly half biologists who picked up the computer bits and half computer scientists who picked up the biology. The first MScs in bioinformatics were just coming in, so that may well have changed now.

Also since you know where you will be living and therefore your target employer do you think you could find someone there doing the line of work you want to get into and ask their advice and for mentoring? Depending on where you are based EMBL offer short courses in various aspects of bioinformatics which might help you decide.

I work in admissions for a CS dept (northern RG). Here's my tuppence-worth!

We do a one year taught masters in IT for non-cs graduates. While we don't ask for a computing degree, we do look for an understanding of programming concepts and terminology. A science background is often found to be useful.

For our masters by research, we look for a strong first degree in CS or a cs-related field, with evidence of good research skills in that degree.

Our UG degrees ask for A level standard Maths (at least a grade B) and this is essential. (Strangely, it's not for our taught masters courses!). We teach the theory and practice of programming as well as the fundamentals of certain languages. This gives students the ability to pick up a language in its most appropriate context (I think we're talking the same thing, pom?)

Other posters know much more about the industry than I do but, from an admissions perspective, I think you would very much suit a taught masters. A year of intensive study (rather than 3 or 4 years at UG) would get you back out into work quicker and more cheaply too.

pombear you talk about learning programming languages as someone who hasn't programmed before. Have you tried teaching yourself in code academy? I have always worked in teams with only software engineers, unlike cdtaylornats. I would say most can program before they begin a CS undergraduate. We all self taught ourselves in our teens. So I don't think a CS degree teaches programming. It however does plug the gaps in our theoretical knowledge.

I think a CS degree will teach you things. And so would a masters. What you will learn I think will depend on your degree course.

I have no idea what MOOC teaches.

I think the best thing to do is still talk with the data analysts in the medical school. The hardest thing is landing your first job! Then you can always upskill by studying part time. I assume you don't have young children which will stop you doing this?

I'm going to be the analyst person in a large medical research group, for example, then I need to know...

that's kind of my job.

A few questions?
Have you done no calculus at all, do differential equations fill you with horror?

Montecarlo analysis, logistic regression, diffusion models, step-wise modelling: what do those words mean to you?

If you struggled with (failed) math before, are you ready to leap into it now? We do a huge amount of data analysis that isn't remotely complicated math, but tonnes to do with research governance, quality checklists, systematic reviews.... Have you done any kind of modelling (in spreadsheets, maybe?)

How committed are you, really, to this career path? Do you want to do genetic sequencing specifically?

What country are you in? I'm not sure if UK options have any relevance to what you can do where you are.

Thankyou everyone.

Aftershock, the mentoring idea is a good one - I will look into it. Bioinformatics, while it would definitely make sense, isn't a viable career path here, as molecular work is incredibly expensive (about 5x the cost in the UK or Europe), so relatively few people (compared to the UK or Europe) have sequence data that needs analyzing. More general analyst jobs for a mix of sequence data, epidemiological data, social science data etc. do come up from time to time.

MarvinKMooney - I completely agree that it's inappropriate for me to be enrolling in a research master's, and a taught master's would be the way to go, if I were anywhere that had one. At the moment there are no relevant taught masters' on offer here, and I can't afford to be an international student anywhere (I'm self-funding).

HeadDreamer - Your point about being able to code before doing a degree is definitely worth considering. Maybe I should really step back and actually develop some coding skills before leaping in, because I definitely haven't self-taught anything particularly useful yet.
For the biological work I've done I've used R a lot, I've done some scripting in bash and perl, I started playing with C, and I've done a small amount of python but have never had to use it for anything since I finished my PhD. When I started in the 1990s all but one of the programs we used for analyses were command-line and had fairly complex syntax (no idea what the language was). So, I'm used to getting syntax right (and typing "man..." and googling stackexchange a lot) but that's only a very small step on the path to actually being able to do anything independently.

MedSchoolRat - these terms mean things to me, I have some idea of when/where they might be used, but I can only describe them at the most basic level - which is why I would want to do maths and stats subjects in an undergrad degree.

Re: lack of track record, I agree that's something I'd want to take seriously. I struggled with maths before at uni because large-scale life events got in the way, so I spent no time on it, and I couldn't blag plausibly after skim-reading a lecture outline, like I could in biology. I didn't ever have the sense that "I couldn't do it" (I did definitely have that feeling with chemistry), it was more that I wasn't keeping up because I was failing to go to lectures or do assignments, but it wasn't particularly difficult when I read through the textbook after I'd finished that year of uni.
Re: sequence data - see above in this post. It would make more sense, but I wouldn't be able to get a job.

I think posts from HeadDreamer and MedSchoolRat make me think I should step away from thoughts of university admissions and spend a few years in part-time work doing something else, while I learn to code and make sure I'm on top of A-level equivalent maths & stats.

Thanks all.