Easy Data Transform v1.40.0 released

Easy Data Transform v1.40.0 is now available for Windows and Mac. Improvements include:

• Run>Run Selected now also processes any items downstream of the currently selected items.
• You can now choose to copy data from the Right pane dataset table either with or without the header

• The Stats transform now has additional Calculation options Percentile (inclusive) and Percentile (exclusive).
• The Row Num transform now has an additional New Column Name option.
• If Run>Auto Run is unchecked Run>Run Unprocessed or Run Selected will now also try to re-run items with errors.
• There is now an additional file name variable {<input alias>.folder}.
• Bug fix for column propagation issue.

Find out more and download the new release from the release page:

Download the update

This is a free upgrade for everyone with a valid v1 license. If you purchased a v1 license you can have it resent here.

If you don’t have a v1 license you can purchase one here.

Thanks to everyone who sent us bug reports and suggestions for improvements. Please let us know if you have any problems with the new release.

New videos

We recently added 2 new videos to the Easy Data Transform Youtube channel:

• How To Convert Excel To Markdown
• How to change CSV file text encoding

Likes and subscribes help us with the YouTube algorithm and are much appreciated.

In the forum

Recent posts in the forum that might be of interest include:

• Strange Date conversion bug
• Replace - replace whole cell value on partial match?
• Filtering a list?
• Merging data dynamically from rows to columns

New newsletter service

We were having problems with the service we use for sending out this newsletter. So we are trying a new provider, mailerlite.com. Please let us know if this causes any issues.

AI hype

Personally, we don’t see large language model (LLM) ‘AIs’ replacing data wrangling tools like Easy Data Transform any time soon, if ever. This is for various reasons, including:

• LLMs are sophsticated Markov chains. They don’t really understand the world in the way a human does.
• LLMs are general purpose, and consequently not optimized for specific tasks like data wrangling. If you have millions of rows of data to process and analyze, you want something specially optimized for that purpose.
• You don’t know how confident you can be in the answer an LLM gives you. If they don’t know the answer, they will just make something up without warning you. There are many examples online of LLMs confidently asserting things that are completely wrong. Or generating plausible looking computer code that gives the wrong answer. This is bad.
• Their user interface is generally natural language/chat, which is notoriously ambiguous. Computer languages, whether text-based or visual, exist for a reason.
• While undeniably impressive, their capabilities are over-hyped by the naive and those out to make a quick buck. This is always the way with new technologies.

Once they have followed the inevitable hype->disillusionment->realism cycle, I am sure they will become a useful tool. Perhaps for advising people how to best use more specialized tools, such as Easy Data Transform?

And finally…

Technological change is nothing new.

Andy Brice
Oryx Digital Ltd
25-Apr-2023