-1
$\begingroup$

When ever the word data science pops up people generally become quick to move to machine learning. Is that the right thing? For a data scientist isn't the handling of data (collection, pre-processing, visualization, etc.,) more important?

I am aware of the thread What is valued more in the data science job market, statistical analysis or data processing?, but the answers really didn't help me and the job market has changed since then!

$\endgroup$
2
$\begingroup$

There’s a lot of statistics that isn’t machine learning: experimental design, inference, interpretable models. All three could be much more important than machine learning, depending on the job.

Then there’s the part that statisticians don’t like, which is that most of what a data scientist does is argue with data sets that are in nasty formats. That could be of considerable importance to a machine learning group, but someone good at programming with no knowledge of predictive modeling could be quite excellent at such a task.

$\endgroup$
9
  • $\begingroup$ So in essence you are saying that every group which focueses on data science activites needs people who are and aren't machine learning focused? $\endgroup$
    – Aymuos
    Aug 27 '20 at 2:39
  • $\begingroup$ Not every group doing data science is going to need bona fide machine learning. There’s still a role for classical techniques and their more modern interpretations (e.g. looking at the credible interval for a mean to do some Bayesian analysis rather than a frequentist confidence interval or t-test). This mostly comes from job hunting a few years ago, but the non-predictive-modeling data science jobs seem most prevalent in biology and social sciences (maybe not finance, if you consider that a social science). $\endgroup$
    – Dave
    Aug 27 '20 at 3:14
  • $\begingroup$ Oh okay! Thank you for the answer :) $\endgroup$
    – Aymuos
    Aug 27 '20 at 3:19
  • $\begingroup$ I appreciate you accepting my answer, but it’s usually preferred that members wait a bit longer before accepting answers. An accepted answer is a signal, potentially, that additional answers are not needed, and someone else might come along and post a better answer. $\endgroup$
    – Dave
    Aug 27 '20 at 3:21
  • $\begingroup$ Oh okay, thank you! I'll keep that in mind for the next time. I never really though about the angle which you presented and hence accepted it! $\endgroup$
    – Aymuos
    Aug 27 '20 at 3:23

Not the answer you're looking for? Browse other questions tagged or ask your own question.