Aunt Pythia’s advice: not at all about sex
Aunt Pythia is yet again gratified to find a few new questions in her inbox this morning. Sad to say, today’s column really has nothing to do with sex, but I hope you’ll enjoy it anyway. And don’t forget:
I’m an academic in a pickle. How do I deal with papers that are years old, that I’m sick of, but that I need to get off my slate and how do I prevent this from happening again? I always want to do the work for the first 75% of the paper and then I get bored. But then I’m left with a pile of papers which, with a biiiit more work, they could be done.
Not Yet Tenured
One thing they never teach you in grad school is how to manage projects, mostly because you only have one project in grad school, which is to learn everything the first two years then do something magical and new the second two years. Even though that plan isn’t what ends up happening, it’s always in the back of your mind. In particular you only really need to focus on one thing, your thesis.
But when you get out into the real world, things change. You have options, and these option make a difference to your career and your happiness (actually your thesis work makes a difference to those things too but again, in grad school you don’t have many options).
You need a process, my friend! You need a way of managing your options. Think about this from the end backwards: after you’re done you want a prioritized list of your projects, which is a way more positive way to deal with things than letting them make you feel guilty or thinking about which ones you can drop without deeper analysis.
Here’s my suggestion, which I’ve done and it honestly helps. Namely, start a spreadsheet of your projects, with a bunch of tailored-to-you columns. Note to non-academics: this works equally well with non-academic projects.
So the first column will be the name of the project, then the year you started it, and then maybe the amount of work til completion, and then maybe the probability of success, and then how much you will like it when it’s done, and then how good it will be for your career, and then how good it will be for other non-career reasons. You can add other columns that are pertinent to your decision. Be sure to include a column that measures how much you actually feel like working on it, which is distinct from how much you’ll like it when it’s done.
All your columns entries should be numbers so we can later make weighted averages. And they should all go up when they get “better”, except time til completion, which goes down when it gets better. And if you have a way to measure one project, be sure to measure all the projects by that metric, even if they mostly score a neutral. So if one project is good for the environment, every project gets an “environment” score.
Next, decide which columns need the most attention – prioritize or weight the attributes instead of the projects for now. This probably means you put lots of weight on the “time til completion” combined with “value towards tenure” for now, especially if you’re running out of time for tenure. How you do this will depend on what resources you have in abundance and what you’re running low on. You might have tenure, and time, and you might be sick of only doing things that are good for your career but that don’t save the environment, in which case your weights on the columns will be totally different.
Finally, take some kind of weighted average of each project’s non-time attributes to get that project’s abstract attractiveness score, and then do something like divide that score by the amount of time til completion or the square root of the time to completion to get an overall “I should really do this” score. If you have two really attractive projects, each scoring 8 on the abstract attractiveness score, and one of them will take 2 weeks to do and the other 4 weeks, then the 2-week guy gets an “I should really do this” of 4, which wins over the other project with an “I should really do this” score of 2.
Actually you probably don’t have to do the math perfectly, or even explicitly. The point is you develop in your head ways by which to measure your own desire to do your projects, as well as how important those projects are to you in external ways. By the end of your exercise you’ll know a bunch more about your projects. You also might do this and disagree with the results. That usually means there’s an attribute you ignored, which you should now add. It’s probably the “how much I feel like doing this” column.
You might not have a perfect system, but you’ll be able to triage into “put onto my calendar now”, “hope to get to”, and “I’ll never finish this, and now I know why”.
Final step: put some stuff onto your calendar in the first category, along with a note to yourself to redo the analysis in a month or two when new projects have come along and you’ve gotten some of this stuff knocked off.
Dear Aunt Pythia,
I am a freshly minted data scientist working in the banking industry. My company doesn’t seem they know what to do with me. Although they are a ginormous company, I am currently their sole “official” data scientist. They are just now developing their ability to work with Big Data, and are far from the capability to work with unstructured, nontraditional data sources. There are, apparently, grand (but vague) plans in the future for me and a future DS team. So far, however, they’ve put me in a predictive analytics group. and have me developing fairly mundane marketing models. They are excited about faster, in-database processes and working with larger (but still structured) data sets, but their philosophy seems to still be very traditional. They want more of the same, but faster. It doesn’t seem like they have a good idea of what data science can bring to the table. And with few resources, fellow data scientists, or much experience in the field (I came from academia), I’m having a hard time distinguishing myself and my work from what their analytics group has been doing for years. How can I make this distinction? And with few resources, what general things can I be doing now to shape the future of data science at my company?
Newly Entrenched With Bankers
First, I appreciate your fake name.
Second, there’s no way you can do your job right now short of becoming a data engineer yourself and starting to hit the unstructured data with mapreduce jobs. That would be hardcore, by the way.
Third, my guess is they hired you either so they could say they had a data scientist, so pure marketing spin, which is 90% likely, or because they really plan on getting a whole team to do data science right, which I put at 1%. The remaining 9% is that they had no idea why they hired you, someone just told them to do it or something.
My advice is to put together a document for them explaining the resources you’d need to actually do something beyond the standard analytics team. Be sure to explain what and why you need those things, including other team members. Be sure and include some promises of what you’d be able to accomplish if you had those things.
Then, before handing over that document, decide whether to deliver it with a threat that you’ll leave the job unless they give you the resources in a reasonable amount of time or not. Chances are you’d have to leave, because chances are they don’t do it.
Please submit your question to Aunt Pythia!