Simon Breton

@martin_casado why is anomaly detection never work?

Systems ideas that sound good but almost never work: - DSLs - Live migrating process state - Anomaly detection - Control loops responding to system load - Multi-master writes - p2p cache sharing - Hybrid parallelism - Being clever vs over-provisioning .. What else?

2-minute survey are usually 15-min of pain survey. This one is more like 5-min survey and not painful at all + the ending really worth it!

Analytics can be a real pain. We’ve got a 2-minute survey to try to put a finer point on what that looks like. Whether your an analyst or you’re simply an end-user of analytics, we’d love your (anonymous) contribution: bit.ly/analytics-pain.

@ThatMoviePage @MonsieurBobine Une video d'une heure sur le contexte de ce qui est presenter dans ce tweet et une enquête sur les intentions du réalisateur?

Thanks to Twitter I was able to get a copy of the 35mm scan of Terminator 2 and I'd forgotten how blue it is in comparison to how it's presented these days.

It's launch day at facts & feelings @less_and_better. Check out what our CEO, @davd_mcbride had to say about it: medium.com/facts-feelings…

A GTM template I shared during @supervveek 🤓 zielinsky.alejand.ro/increasing-the… Among many things it does: -Contains all of the recommended GA4 Events & Parameters -Pull event data directlly from a dataLayer.push() -Add custom parameters and user properties -Load GTAG from a custom URL

.@supervveek AI Roles and Risk presentation slideshare.net/slideshows/ai-… #spwk

25 days before @supervveek! The unique international conference of #analytics Just for fun, I built a report with geo data of the speakers & 2 recent #LookerStudio features: A connection map & a transposed table as a pictures ribbon 😁 The report: bit.ly/3TNTajD #spwk

🙏 Huge thanks to all our attendees, sponsors, and organizing committee for making Measure Camp Warsaw an incredible event! 🚀 Your contributions and dedication helped create a memorable experience. Let's keep the data discussions going! 📈 #MCWarsaw

@neelesh_salian I think timing is key. When will it be noticed? after a couple of days? a week? one month? three months? You can have valuable data set activated only a couple of time every year. Impact is when people notice. obs is different.

If an ETL fails inside an organization and there’s no observability, did it really have an impact?

@thyng i will need to ask for a couple more days off for this one!

WooHoo

@matsonj i spend time over the last couple of week refactoring all the cloud functions i'm using for a client as ETL to make sure i can properly log everything and especially catch all "possible" errors.

today in "data engineers don't look at the data" - the api calls were being made successfully but returning a blank response, ergo the "job passed" (green in airflow).

2nd Ticket Release for first Measure Camp Warsaw today at 12:00!

That's very clever actually, use PK/FK to optimise SQL Queries in #BigQuery

@fred_pike @GeoffBCampbell @googleanalytics @geoff and you should not worry about consent mode blocking anything 🤣even when consent is denied it is still collecting data!

@GeoffBCampbell @googleanalytics Hey @Geoff - I've not had any problems with OneTrust and GA4. FWIW, I'm not using consent mode; I use the blocking triggers. I fire the GA4 config tag on oneTrustGroupsUpdated or on historyChange and it has generally worked quite well.

@GeoffBCampbell @googleanalytics Did you check the network tab? or the intra_day table? maybe this is just an issue with the DebugMode.

Evaluating the data quality of a pipeline is complicated. There's business value to consider. There's maintenance. There's ROI I wrote a new newsletter covering all these topics. If you go to my profile you can catch it since Elon would downrank it otherwise!

@AnalyticsNinja Be careful for what you wish for

For my birthday, I would like to have a high performance liquid chromotograph to measure the scoville units of food I eat. Right now I am eating some sauce I made that I would wager is 600000 scu, but I don't know for sure. It's damn hot though (and good on chicken)

.@dataform The number of model that can be displayed per folder from the GCP interface seems to be limited (ie if you have too much models the one at the bottom of the pile doesn't show up). Can you confirm this behaviour?

Just posted on the @rittmananalytic blog : "ChatGPT-Powered Data Analysis using Cube, Delphi and the Code Interpreter Plugin" blog.rittmananalytics.com/chatgpt-powere…