Dimodelo

Dimodelo publishes information, frameworks, courses and tools for aspiring data professionals.

#datawarehouse #dataengineering #datalakehouse #etl

[Dimodelo @dimodelo]

Lesson 4 of 8, from the full case study: https://t.co/w1DRWgSlBl

[Dimodelo @dimodelo]

Query history captures what people did, not what the model means. Every shortcut in every one-off and prototype analysis.

[Dimodelo @dimodelo]

Lesson 3 of 8, from the full case study: https://t.co/w1DRWgSlBl

[Dimodelo @dimodelo]

Lesson 3 from handing an AI analytics agent a complex request with no context:Context matters more than ever.Our table and column descriptions let the AI find a table the request never https://t.co/lPFcKZfTul also picked a DEV integration schema, because nothing told it not to.

[Dimodelo @dimodelo]

Lesson 2 of 8, from the full case study: https://t.co/w1DRWgSlBl

[Dimodelo @dimodelo]

Four paths between two tables. The AI chose one, short-circuited the join, and under-reported.

[Dimodelo @dimodelo]

Lesson 1 of 8, from the full case study: https://t.co/w1DRWgSlBl

[Dimodelo @dimodelo]

Lesson 1 from handing an AI analytics agent a complex request with no context:

It's a good start. But it would take someone who understands the model in greater detail than the AI does, and has better judgment, to reach the right outcome.

An analyst's knowledge still matters.

[Dimodelo @dimodelo]

What it got right, what it got wrong, and why (queries included): https://t.co/w1DRWgSlBl

[Dimodelo @dimodelo]

We handed a complex analytics request, verbatim, to Databricks Genie with almost no context.

Right: the right tables, the right filters, even the "Active" flag.

Wrong: under-reported, joined by name despite a foreign key, mixed prod and dev data.

Plausible. Subtly wrong.

[Dimodelo @dimodelo]

@matt_barrie This is so dumb. Especially when you consider burning gas to cook food, is nearly 100% efficient (i.e. 100% heat energy), vs running a gas turbine to generate electricity is, on average, 30-40% efficient, and you lose another 10% on transmission across the wires.

[Dimodelo @dimodelo]

@azamsharp @matsonj holy crap. 😭

[Jacob Matson @matsonj]

Do not do this.

[Dimodelo @dimodelo]

Developers often think of a whole Dimension as being Type 1, 2, etc. That is a mistake. A Dimension can have Type 1, 2, 3 etc. attributes at the same time.

https://t.co/4RzAyyPwOz

[Dimodelo @dimodelo]

Is it slowly changing dimensions or slowly changing attributes? ... Read this deep dive into slowly changing dimensions.

https://t.co/4RzAyyPwOz https://t.co/4RzAyyPwOz

[Dimodelo @dimodelo]

@ergestx I like this take. Low variation processes = better data quality.

[Dimodelo @dimodelo]

What is a Star Schema (and why it’s important)

https://t.co/McaARKhHYe

[Dimodelo @dimodelo]

@sweatystartup But the phone is already paid for - by the consumer. It’s equivalent of a person paying for a storage unit, and then every delivery to said storage unit is clipped 30%. Fair enough if you also supply the delivery service, but not, if you say, you can ONLY use my delivery service.

[Dimodelo @dimodelo]

@bigyahu I got an email from my accountant to tell me I had an email from myGov, but not to worry, it was nothing... and I probably paid for the privilege!

[Mike Rabinovici @mikerabinovici]

After the launch of Dimodelo Data Warehouse Studio for Azure Data Warehouse a couple of weeks ago, we've had a lot of questions about the various features our development team built in. Here is the list-https://t.co/H5S5LXmapX #dimodelodatawarehousestudio #AzureSynapseAnalytics