I'll both agree and disagree with you here. I agree in the sense that the tools I find most useful are working closely in the REPL by applying specific filters and transformations and looking at the text outputs. I think I am less excited about LLM interpretation, but it's certainly increasingly useful. If you're trying to get something done, it's probably usually best to get smart with textual outputs.

That said, there's a gap in that stack that I think sonification fills. The gap is making a case to others. When you want to get your point across on social edia or to your boss, a fancy visualization goes a long way. While sonification is kind of not really there yet as a toolset, I think sonification can fill the role of something that can be shared with others to make them understand some point you're making. Sonification does well on social media, and while it's probably not going to dominate boardroom reports like a pie chart, I could see sending a file to your boss here and there in specific situations.

I do think the technology needs to get a lot easier to use, and to work on a wider variety of hardware.

Patrick Smyth, PhD

Chief Learner

Iota School

GitHub | Twitter | LinkedIn

On Sun, Mar 31, 2024, at 10:19 AM, Pietro Sanchini wrote:

Hi Gabriele,

I followed the course with Sarah, but I confess that I am still not convinced of the usefulness of this type of tool in the daily life of a blind data scientist. Sometimes we have data with a lot of noise, biases, and important details that will be difficult or, in my opinion, impossible to capture completely with sonifications.

It can be useful and even somewhat playful for astronomy, but as a data scientist who works in industry analyzing and applying predictive models in big data and business problems, I don't see much use for it.

That's why I believe that it is much better to invest in two fronts: LLMs for graph description and statistical tools for data analysis - hypothesis testing, in-depth study of linear algebra and probability, study of numerical analysis, etc. I have been investing in the second front for four years now and although I still have a lot to learn, I feel increasingly confident in working with data without using graphs.

Again, this is just my personal opinion and there may be those who disagree.

Kind regards,

Pietro

My first question has to do with a subject I care a lot, sonification.

I tried astronify but I'm not very happy with it. In my opinion it lacks

many things like stereo capability, using different waveforms, sonify

more than 1 serie of data at the same time and a usless complex grammar.

I would like to try something more complete and maybe easier to use.

Do you know something like that, to import as a python library?

Gabriele Battaglia (IZ4APU)

--... ...-- -.. . .. --.. ....- .- .--. ..- - ..- . .

Sent from my Giant Desktop PC

_______________________________________________

Data Science mailing list -- datascience@blindcoders.com

To unsubscribe send an email to datascience-leave@blindcoders.com

_______________________________________________

Data Science mailing list -- datascience@blindcoders.com

To unsubscribe send an email to datascience-leave@blindcoders.com