Should I study data science at university?
I am a generalist who is good with numbers and words. I am not sure if I should pursue an economics and politics degree or do data science instead as it is more practical and there's a higher chance that I could get a job at the end of it. #careeradvice # #first-job #job-search
3 answers
Mark’s Answer
Another career to consider is Actuarial Science. beanactuary.org has more info.
Hi Mark, why did you suggest this student look into Actuarial Science? Do you think it would be good fit for them based on their interest?
Gurpreet Lally, Admin
Danielle’s Answer
Hi Ben,
It's good you're considering the application of your major, how general/specialized it is in order to secure jobs.
If you're not in college yet: research schools with those programs, see which ones sound like a good fit for you.
If you're at college: Go talk to advisors in those departments/majors. You can see if they have any post-graduation data regarding the alumni from that major, what jobs they do. That information might help you evaluate what works best as far as your interests.
Also, take a look at O*Net, it's a resource that shows what kinds of preparation is needed for specific jobs. You might find out about careers you hadn't considered/don't know about yet.
Danielle recommends the following next steps:
Bob’s Answer
This is a new answer to an old question, sorry about that! Maybe it will help someone else along the way who has a similar question.
Data Science, as a term is similar to the term "Jazz". When someone asks you if you like Jazz are they asking about Smooth Jazz, Beebop Jazz, Dixieland Jazz, Cool Jazz? Essentially, the term Jazz is a term that denotes a certain skillset in a discipline, but doesn't specifically isolate the application you will ultimately leverage with that specific skill set. And Data Science means something similar. You typically understand that you are building business intelligence, but you will typically be predicting things, but what exactly will you predict? Will it be mortality rates, automobile driving patterns, whether or not life exists on Mars, if a person will get cancer, you get the point.
Typically Data Science will marry three main skills:
1. Computer Science skills
2. Math & Stats skills
3. Subject Matter expertise (credit card fraud is its own field for example)
The golden unicorn is someone who has all three skills in a given discipline, and that is really difficult to do as a subject matter expert in a given area may be a PhD. For example, Andrew Ng, a Data Science pioneer has a course in which they used Neural Networks to help a blind man see, or they are helping to build AI driving technology. Last I saw with Andrew he was applying his AI skills in the medical field mostly. So, as a data scientist, understanding the problem you NEED to solve, then understanding if you've solved it, can often take specific understanding of the subject matter, as is the case with the medical profession.
You may also benefit from knowing the typical workflow of a data scientist:
1. Identify the problem
- Means, identifing what the objective of the solution could be.
- Also identifying a set of questions that will help you build the correct data set
2. Acquire the data
- Obtain the right data set
3. Parse the data
- Explore the data. Perform some quality checks on the data.
4. Mine the data
- Determine the sampling methodology for the sample data.
- Format the data
5. Refine the data
- Identify trends and outliers
- Document the data and transform it. Sometimes this means building intermediate data sets to help speed up the final solution.
6. Build a the final data model
7. Present the results
Data scientists typically work in teams. Usually, there will be people who perform the following types of tasks:
- Data visualization
- Building machine learning models
- People who can communicate the outcome with confidence and clarity
- Domain experts
- Programming
- Building statistics models
You may also want to know that you are typically dealing with unstructured data. This is data that often has errors in it. So, you must be able to detect the errors and resolve or remove the data in a way that is explainable as the outcome of this action may materially effect the confidence of the prediction.
The field can be rather rewarding if you work on topics that you find interesting.
Hopefully this helps you or someone else.