hierarchical tables?

Question

I have a shallow tree to display: action -> action item -> error

There can be millions of actions, tens to thousands of items per action and (usually) 0 to 1 errors per item. There can be also errors for an action (then I guess tens of them).

Obviusly with that amount of data actions and action items need to have some filtering.

The goal is to:

• quickly see the status of an action
• to browse errors for an action
• to see on which action items the action was executed

All of these 3 object have more properties than the columns I could fit resonably in a table (but not too much, 5-15 at the moment).

Do you have some pattern for showing such data? Would you use 2 paginated tables for actions and items, the items table depending on actions table? Some kind of paginated tree? What is your suggestion for such kind of data?

Answer 1

This is difficult to answer without more context.
What are the types of actions, items, and errors? What function/s are you trying to perform or what problem/s you are trying to solve?

It looks like you are wanting to show potentialy thousands of millions of entries. Trying to show this quantity of data in a meaningful way is pointless without specialist visualization methods/tools such aggregate vizualizations or treemaps.

For an idea of just how pointless it is trying to use something like tables for this try doing a google search for 'music' - out of the 8,470,000,000 results you are only shown the first 10-100 items (depending on settings). Even at 50 results I would consider this to be cognitive overload.

Tring to show this much data you will run into some of the following problems:

• clutter
• performance issues
• information loss

The Visual Information Seeking Mantra states:

Overview First, Zoom and Filter, Details-on-Demand
- Ben Shneiderman

This was evolved to the Visual Analytics Mantra:

Analyze First, Show the Important, Zoom, Filter and Analyze Further, Details-on-Demand
- Daniel Keim

What these break down into is that sometimes you need to use the power of the machine first and let it discover if there is anything interesting/useful to show before visualization is used.

See also Fell in Love with Data and Extreme Visualization: Squeezing a Billion Records into a Million Pixels