The Time Series Refinery : a new major Version

It's been a while since we gave news - we were busy building the next iteration and packing new productivity tools into the refinery !

We now have a news page describing the full menu. Here you will find a roadmap and the contents of current and past versions. We expect on average two major versions per year.

Let's focus on the most important item of the latest release : the find API method and the findseries formula operator.

Find takes a string of filter expression in a small Lisp-like language providing a number of filter predicates.

We can filter on names (a good naming convention is worth its weight in gold !) and of course metadata, including with inequalities.

Lastly we can compose a query filter and give it to findseries and then use that value to feed an add operator. Such a dynamic way to build expression based on names / metadata can certainly help build complex formulas and evolve them by just playing with the series metadata.

Persistent named queries can be managed using the new basket API.

We hope you find this as pleasing as we do.

On the UI side of things, I'd like to show two interesting novelties.

First in the tsinfo view (that gives a identity card of a series), we made it possible to incrementally expand the formulas using a slider. Like this :

It will be now very easy to perform a downward audit of your big formulas, especially as they become very deep (kudos to our friend that wrote a formula of depth 24 and with over 1.500 members ! We will keep his name a secret :).

Oh, and all the series names are now directly clickable. This is very handy.

And now, for something completely different: the new filters on the tasks manager task list.

It will be a breeze to find failed tasks for a given input. Finding those quickly is a core part of a data management specialist duties.

There are a number of other quality-of-life enhancements that you will discover on duty. We hope you enjoy them as we do !

Best regards,

The Pythonian Team.