Trust In SODA’s Post

Transcript

Being in that kind of environment where everything is manual and you have to do everything yourself. Like obviously, as you said, time has moved on dramatically since, you know, doing maybe they haven't, but like generally speaking, people have. Surely that that has kind of influenced the way that you look at how the systems that you work on today and and then go, well, it shouldn't do that because we just need to be hyper efficient. Yeah, yeah. I mean, it's interesting I. I always think of data in terms of of. Who's touched it and where is it coming from? Right. Like so, so I, I think that like, like I said, I'm always sort of assuming there is a human somewhere inputting data incorrectly somewhere because I've, I've been that person on myself. So I think the influence that it has and you know, I think what's really. Interesting about, you know, anything when you're scaling data analysis, right? So that includes cleaning with, you know, your SQL or your pythons of the world and then doing something like machine learning or what AI is doing now. I'm always thinking back to OK, where is this gonna get messed up? So when you are doing your data cleaning, I think the way that I see data is like. What are those users? If it's user inputted, like where are they commonly gonna go wrong so that we can code around it in a, in a way that's really consistent. And, and, and when you understand those patterns of data errors, then then that's like the best pairing of code cleaning with a really, you know, that's kind of a generic term like messy data, but I always think of it in terms of behavior of like, you know. What Yeah, what is common about those errors that people make? And, and we see this, right with like things like typos, like often people tend to mess up the same words. And those are things that you can you, you can data collect on and, and mentally and then take note of. So when you're doing these, these more sophisticated data cleanings, you can do it at scale and do it really efficiently.