UNIFORM LA Japanese BDU Camo Cargo Pants Drop, going on right now.
Uniform LA's Japanese BDU Camo Cargo Pants are now live. These cargos are based off vintage US Army BDU (Battle Dress Uniform) cargos. They're made of a premium 13.5-ounce Japanese twill that has been sulfur dyed for a vintage look. Every detail has been carried over from the inspiration and elevated. Available in two colorways, tundra and woodland. Please find them here
Good luck!.
STYLE. COMMUNITY. GREAT CLOTHING.
Bored of counting likes on social networks? At Styleforum, you’ll find rousing discussions that go beyond strings of emojis.
Click Here to join Styleforum's thousands of style enthusiasts today!
Styleforum is supported in part by commission earning affiliate links sitewide. Please support us by using them. You may learn more here.
I think we’re saying the same thing. The type of model is fundamentally unsound for the task. They are fine analytical tools, but not good predictive ones. Sometimes the term “highly predictive” in scientific and technical literature is misread. It often means “models the non training data well”, not “Is able to be extrapolated into the future.” I think that this is the problem with laymen reading any type of literature meant for mature researchers. Not unlike law, the words have specific meanings but are close enough that the meaning can be easily misinterpreted.Isn't the slur word used "data analyst"? Now the problom might be more constitutive than that: past data models might not have a strong fashion predictive capability no matter who builds them. Maybe at best it can be an ok tool as part of a much larger toolkit.
I think that we’re talking towards the same point. I’ve seen a lot of business models project YoY growth that is either unsustainable or simply not feasible, and not just in fashion. At the end of the day, there are only so many people living so many hours with so much space and so much money. Physics gets in the way.In my ~decade of working in/around commerce start-ups, I struggle to think of single data science team that was really moving the needle for their business, but plenty that needed to exist to justify the tech multiple.
Sympathetic to @Fuuma 's point here - if anything, the most common planning disasters I've seen are where someone takes a flimsy-but-inoffensive model, tie wraps it to an obscene YoY growth projection, and places a PO w/o ever encountering the fact that they're now ordering more units of something than there could ever possibly move. But when everyone's wanting to see a big number, and people are playing cashflow games with inventory revolvers...
We're indeed, roughly speaking, saying the same thing. I'd add a few points:I think we’re saying the same thing. The type of model is fundamentally unsound for the task. They are fine analytical tools, but not good predictive ones. Sometimes the term “highly predictive” in scientific and technical literature is misread. It often means “models the non training data well”, not “Is able to be extrapolated into the future.” I think that this is the problem with laymen reading any type of literature meant for mature researchers. Not unlike law, the words have specific meanings but are close enough that the meaning can be easily misinterpreted.
I would say that there are two separate issues here:We're indeed, roughly speaking, saying the same thing. I'd add a few points:
- There is a debate as to how useful various models are at, as you said, "extrapolating the future" so this question is not settled. We're probably two people more on the "never/almost never" works (works could obviously be heavily qualified) end of the spectrum (I'm not a data scientist though so being at the very least agnostic is easy) but even critical data scientists may be situated toward the other end of the spectrum.
For these models in general, yes. My friend Cathy O'Neil wrote a very good book which is mostly about this: "Weapons of Math Destruction." She is a mathematician, so you have to forgive her for the title. https://www.abebooks.com/servlet/BookDetailsPL?bi=31090427161
- When we say that the models do not work we are dismissing unstated goals. One of the often left unmentioned objectives of the bourgeois hegemonic bloc is social reproduction (of the bourgeois class). Turning past data into future reality is one of many great tools to accomplish that goal (people stay in the lane they were born in etc.). This isn't some Soros sat with Bill Gates type conspiracy where everyone intended for this to happen and there's a dated memo.
Sure. Fashion cycles too fast (though there is a long tail for adoption) and there is way too much qualitative data that is extremely difficult to quantify, in general, compared to many other categories. For example, trends in large appliances or on flooring are far longer and more predictable.
- Might there be some specifics having to do with fashion that make models even less reliable as analysis or extrapolating tools than in other fields?
I really like Amanda Mull’s articles.America Is Drowning in Packages
UPS workers have an impossible job in the Amazon age.www.theatlantic.com
I used up free articles so this is blocked for me.
It’s also a not more fun. TikTok and streaming stars are the new frontiers of influencer entertainment and the users are way more connected with the creators than say, YouTube or Instagram.
Kind of interesting. I feel a bit skeptical that they could really capture an exact look using thrift store finds, and it seems like a lot of it wouldn't end up fitting right, but it's way better than people buying from Shein.