What's Up Wednesday (1-March-2023): Transformer Updating, Building Pipelines with FME, Yet More OpenAI, and Analyzing Driving Data.

Hello FME friends,

I know. It's Thursday. It might even be Friday for you. But whatever day it is, this is still your weekly review of all news FME.

Community Cafe Update

Just a quick mention that I upgraded the community FME Cloud machine to a new FME version, and tightened up a few security settings. So, if you find any problems occur when redeeming webinar badges, do let me know.

And don't forget, you can now get a transcript of which webinars you've attended, and view the full list of available FME Community badges:

Auto-Updating Transformers

The rule is, every FME transformer you place has a version number and stays at that level, even if you upgrade FME itself. You have to manually update a transformer to a newer version:

But there is a saying that every rule is meant to be broken, and that is true here. Some transformers will update themselves automatically.

It's rare, but let's take a quick look at why and how this happens.

This WorkspaceRunner transformer here is version 3:

I've opened it in FME 2022, where the latest transformer version is v4. So far, it hasn't changed. And if I run the workspace, it'll run as a v3 transformer.

But, if I open the parameters dialog and then click OK:

...now it's version 4. It has updated itself automatically!

Why? Usually, it's because of a change to the parameters dialog.

For example, the (Database)Joiner transformer once had a wizard to set parameters. When it changed to a single dialog, the wizard no longer existed. So opening an old Joiner would have auto-updated it in the same way as above, to handle the change.

I mention this concept because we talked about this in our support team, and not everyone realized this could happen. So now you know, and you don't have to worry if you notice an unexpected version update.

ChatGPT: The Gift That Keeps on Giving

If you aren't familiar with that phrase, it means something that is good and just keeps providing more and more benefits. Kind of like FME, really.

Anyway, as if FME+ChatGPT wasn't an impressive combination already, we have another great use, thanks to Oliver Morris at Avineon-Tensing.

What Oliver has created is an FME Server app (running during UK business hours) that reads a PDF file and uses OpenAI to generate a summary of it. Here is me uploading a file:

Here's the main summary:

What's even more impressive, is that it works regardless of size and language, and also extracts keywords and geographical areas found within the text:

I'm impressed that it picked Slovenian out of there. It did miss at least one location (Ljubljana), and it did time-out a couple of times, but that's more likely because the OpenAI API was slow to respond.

Great work, Oliver. I look forward to the Hub transformer that you mention.

DataBuilders Webinars

If you missed the webinar by DataBuilders in Ireland, on the subject of Data Pipelines with FME, then you can find the recording online - and embedded here if you're reading this directly on LinkedIn:

It covers the path to becoming a data-driven organization, and cutting down the time from data capture to data insights. Triple-certified Gavin Park demonstrates loading data into Snowflake, how to run tools using an online app, and how to automate processes with FME Server automations.

What I really like is that you see a lot of the data warehouse (Snowflake) and analysis software (Sisense) and how it fits into the whole process. It's worth a watch.

And there is another webinar soon about Data Quality and FME.

Handling Car Mileage Data with FME

Check out this user story by Miso in the UK. It's all about analyzing how many miles are driven by British cars.

FME plays a great role in

• Validating the source data (no, that car did not drive a billion miles!)
• Standardizing the schema (I think it was CSVs with inconsistent headers)
• Expanding abbreviations (like HY = Hybrid, DI = Diesel)
• Filtering data by attribute to get the required information
• Aggregating data and loading it into Postgres

There were 140 million records to process, and it resulted in this report, which you can download to get the full set of results.

Remind me, what was the FME 2022 slogan again?

That's the one!

Other News

• Here's an upcoming webinar from con terra in Germany, about FME use in municipalities.
• Here is the FME Hub digest for February.
• And here's a list of training courses by Veremes in France.
• Finally, a nice blog post from Bill Dollins on (spatial) leadership.