XGboost, LGBM, pmdarima, stanpy (for bayesian modelling). Plus a few others.
Don't ask me what they do with all of these, I'm just the guy who make sure the forecast keeps being reproducible.
2,752 karma · joined April 14, 2014
Don't ask me what they do with all of these, I'm just the guy who make sure the forecast keeps being reproducible.
Airbyte, looks "okay by the cover". If I want to go any further I would need to see what happens when I need to extract data from an esoteric, unsupported, undocumented, Java-based ERP that is no longer sold by its own developer. Unfortunately, I don't have the time nor the skills to do it. So, in the end, we let our clients write their own ETLs on their own stuff and match their datasets to our data model.
Unless I make the happy uncounter with a tool that focus on these forgotten ERP and CRM and don't completely break down when it encounters a weird FTP server serving "encrypted zip files", I'll just let it go.
Having a "Facebook ads" connector is not enough, and not where problems lies.
> bidirectional data connectors which aim to solve automated ETL (eg moving data from an API to Snowflake), reverse ETL (we pipe the data straight to your CRM)
I need exactly this at work. However, my requirements are :
- it has to be "self-hosted", usually in my clients' VPC, sending data to a third party is a big no-no.
- most of them use SAP, Salesforces and many other esoteric CRMs & ERPs. It has to work for at least the most important ones.
- the target medium is usually the cloud provider's object storage (S3, ...), snowflake is quite rare. It also have to be flexible enough to support our own versioning strategy.
Would it work with your product ?
This sentence irks me on so many levels...
"Voting with a wallet" doesn't work, doesn't exist and is never taken into account anywhere, except in the mind of the American Consumer. Not buying a Remarkable when the subscription was added means your opinion on the matter just never showed up during product decisions, planning or analysis.
Instead, it got changed because Remarkable listen to what its users say and EXISTING users actively disliked that subscription. It has absolutely NOTHING to do with potential buyers avoiding it.
Returns, on that matter, are active users expressing their opinions. Not imaginary buyers "voting with their wallet".
Next in line is how to version your data so you can create consistent "snapshots" at given point in time with no leakages.
Maybe in the US, but nobody is going to question it in the UK, France, Germany, and the rest of most European countries.
If you wish for Adobe Scan to store the scan in "Files" automatically, well, you can't. You have to "share" them individually FROM THE APP to "Files".
And I think you will understand, now : this is the same shit show when trying to upload something via FTP. Like family photos.
The UX is stupid. The workflow is abysmal. The discovery is inexistant. The teams responsible for this at Apple don't care, as it's been like this for _years_.
What else are you going to use an iPad for, except for editing family photos and annotating pdfs ?
I use mine to watch Netflix.
GP was right, iPadOS is never going to work for anyone having to send a pdf to someone else.
And then I tried sending a pdf by email. Oh well.
Can I share it on the corporate FTP ? Oh no.
Let me airdrop this to you... Wait... is this a windows / android device ?
...
Alright, I'll just use my laptop. My iPad is too niche for this workflow.
If your issue is "atomics are slowing down my app", then I assume you already milked the code to the latest micro second of performances everywhere else. This is likely not the case here and not a general advice I would give to anyone.
Remember, some dev in python where concurrency is inexistent, start-up time is horrendous and performances are abysmal compared to rust. (this is exaggerated: of course you can run stuff in parallel in python)
What do you mean ?
If you refer to heap allocations, then you can use Box, Arc, Rc. They are not a "garbage collector" nor do they incur performance hits other than a regular heap allocation.
AKA, a cheap Chinese knock-off.
Truffles, like any other mushrooms, are pretty much a gift of nature. They are very sensitive to climate change, pesticides and other forms of pollution.
Be conscious about it the next time you enjoy a truffled dish in a restaurant.
So… Apparently not…
The USA is probably the worst place in the world for 1) high speed trains 2) buses. And the only place I know where the train _waits_ for cars to go through.
If only they could see by themselves how bad it is during their next trip to Switzerland, the Netherlands, Japan, London, Paris (even France in general) and many many other countries and cities that have a functional network of high speed trains, metro, tramway and regional lines…
I hope you’re not ready to die on that hill.
Europe is not a single country, anyway, here's what happen in France:
Firing is tedious, but not impossible. Hiring people is easy, but you enter into competition with higher salaries - like the US - for remote jobs.
> Firing easily and hiring easily are two sides of the same coin.
Not exactly, they diverge when firing is made more difficult.
Just so you know (and stop using that phrase), there is no way scientifically or otherwise, to attribute the effect of non-sales to a feature, a service or anything else.
Your opinion on the matter just never shows up during product decisions, planning or analysis.
Your "vote" just doesn't exist.
Finding out how to attribute a sale to specific features or marketing is definitively possible and common usage.
Finding out why customers churn and what levers could prevent that is also a very common practice.
But finding out why someone _didn't_ buy something in the first place because of a _missing_ feature is just plain and simply impossible.
If you disagree and can prove that I'm wrong, please send me the maths.
"voting with money" or "voting with your wallet" is a complete non-sense. There is no -absolutely no- way for anyone in the industry to attribute a lost sale to a "missing" feature.
I think this also a symptoms on how Americans view the world through means of consumption. Everything HAS TO be consumed one way or another.
Tim Cook: (40:20) Now let’s talk about iPhone. iPhone has forever changed the world. Every single day people rely on iPhone, from their most demanding tasks to capturing the moments of their lives, streaming their favorite TV shows, playing their favorite games, staying up to date with the news of the world and connecting with friends and loved ones.
Tim Cook: (40:42) This is what drives us to create the best iPhone possible. To create an experience unlike any other, with legendary ease of use, beautiful and durable designs with water resistance and great battery life, with industry leading performance and the world’s most advanced camera systems, and with privacy built in. People love iPhone, and we keep making iPhone better, more powerful, more capable, and even more fun to use. Today, we are thrilled to introduce our next generation of iPhone. Here it is.
Tim Cook: (41:54) [inaudible 00:41:54]
Tim Cook: (41:55) This is iPhone {15,16}. To tell you more, here’s Kianne.
Cayenne: (42:24) Let’s take a closer look at, iPhone {15,16}, starting with design. It has the sleek, flat edge design that people love. It also have a fantastic new charging port, featuring the same USB-C connector as our beloved iPad, developed in a tremendous effort to {fight against climate change and electronic waste,further enhance the connectivity of the iPhone {15,16}}.
From what I can see, it works with 100% of the world's websites.
- loading thumbnails should be near instantaneous for the whole album. If it is taking longer, then let me downscale further.
- loading full-scale, several megabytes, photos must continue in the background. otherwise I’ll be waiting on a black screen for several seconds during a slideshow. It’s something dumb to do, but somehow Plex can’t seem to handle that properly…