Hacker News Reader

Comment by fiatjaf | Hacker News Reader

I don't understand it. When you said "multidimensional" I thought of it as a database that could be displayed as different spreadsheets from different viewpoints, but in the examples I cannot see the multidimensionality, nor where is the data coming from on each calculation, I don't see formulas anywhere also; and I can edit the values manually.

I am probably missing something.

View on HN

Multidimensional in this case refers to OLAP cubes. In other words, it's a relational database with a star schema.

The formulas in this case are built in the "metadata editor". The expectation is that a data scientist would build the tables, and then pass it off as an analysis tool to the MBA types. It can be really unintuitive if you've never seen these types of dashboards before.

Speaking of OLAP, is that the old Microsoft OLAP demo database data I see, the one with all of the bicycle shops?

Thank you.

Considering this (and my experience in the demo), this is probably not the nice tool I was imagining.

Don't know about the specific implementation, but OLAP does not necessarily mean relational with a star schema. There is a distinction between ROLAP (Relational OLAP, which is what you describe, plus snowflakes), MOLAP (Multidimensional OLAP, which is true multidimensional, as in Essbase or TM1) and HOLAP (Hybrid OLAP, which is a combination of ROLAP and MOLAP as in SQL Server Analytic Services.)

Multidimensional means every cell of a worksheet is associated with multiple dimensions like year, month, company, currency and so on. This implementation is not really a ROLAP. Instead of a relational database I use a document database for storing metadata. You can access formulas via context menu on cells (grey cells have formulas on them, white are input). The whole thing has become a bit complex. Some information is in the docs.

Reply on news.ycombinator.com