alt Hacker NewsSure. There are many examples of data scientists attempting to use complex Machine Learning and Deep Learning models to predict machine (bearings, gears, etc.) failures from vibration data, where a simple Fourier Transform (FFT) provides a lot more insight and predictive powers about the same problem.
However, spectrum analysis is not something that data scientists learn at school, yet every mechanical/electronics engineer working in the field knows about it. So, without an expertise in a particular field, data scientists often reach for a big hammer, when more specialized tools exist and are known to the experts in the field.
Replies
It's much worse than that. If you dare to ask that a team speak with the problem owners - mechanics, managers, etc, you will get booted right quick.
Since the 2010's data science has gone from scientific based curiosity in solving problems to pure technicians work. There's a set of algorithms they follow, no exceptions allowed. Kaggle is a horrible anti-pattern.
NB: I am a data scientist.
huh. I'm a professional data scientist, and my masters was in signal processing. In one class the final exam required us to transcribe fourier transforms of speech into the actual words. In another the final exam required us to perform 2d FFTs in our head.
Please be careful about generalizing.
I agree that many 'data science' programs don't teach these skills, and you certainly have evidence behind your assertation.
Yet I suspect that mechanical engineers are not writing software for companies in the large. There are software developers doing so.
I suspect that they should be consulted by data science folks as domain experts.
That said won’t AI replace both? ;)
i confess, i've read both of your comments on this - your analogy and a deeper explanation of the analogy - and i still have no idea what you are saying. i'm not stupid. so first, my feedback here is, it sounds like you are an educator or in an education-adjacent role, and you should focus on making more sense haha. like lay out your beefs clearly, it sounds like you have a beef with interdisciplinary work, specifically between some STEM departments and especially with humanities and STEM departments, which is subjective. you don't have to be objective about everything. you can just say, "i don't like this design thinking thing because i don't like the people involved" or whatever. but i don't know! i cannot figure out what you are saying.
it sounds like your point is: "some ways of solving problems are superior to others." i've heard this take a million times. one perspective i'll offer to you is, math is not the only way to solve problems. it's not even the best way in many cases. not everything can be solved by defining a narrow goal, and then having a dispute about the methods, and then picking some objective method and then applying it very optimally, or whatever. this is also on you, as an educator, to understand! i could give a bajillion specific examples.
but first, you have to concede: an analogy nobody understands is bad, and you have to own that, and two, it's not really clear, what exactly is your dispute with Design Thinking? it doesn't have anything to do with user interfaces... so why the hell are you talking about it? why "Design Thinking people"? What is your beef here?
Another classic example is data scientists trying to model biological processes (or answer questions about processes while ignorant of which components regulate others). Systems biology has a long history of largely clueless attempts to predict outcomes from complex processes that no one understands well enough to model usefully. The biologists know this but the data scientists do not.