We’re delighted with the response to our webinar, Big Analytics for Python Users Without Big Hassles. We received many questions through the chat window. We answer some of them here.
- Do I need to know Array Functional Language (AFL) and Array Query Language (AQL) to use SciDB-Py?
AFL and AQL are the native query languages baked into SciDB. AQL is declarative, roughly analogous to SQL. AFL is a functional language that uses composition of functions to create increasingly complex queries. In practice, you need some familiarity with AFL or AQL. The current (beta) version of SciDB-Py does a pretty good job of limiting the amount of AFL and AQL that you will need to use, and as we move forward that will improve further.
- The webinar showed some graphic visualizations of SciDB data. Where was this rendering performed—on the cloud or on a local machine?
The heat map was rendered on a local machine using the raster option in matplotlib; nothing noteworthy there. But the interpolation to produce the data behind that image was produced on the SciDB side. Remember, the raw data (the NASA MODIS data) is sparse, so we needed to interpolate averages to produce an intelligible image.
- How different is a SciDB array from a SQL table?
There are some similarities and some differences. The coordinate axes (aka “dimensions”) in a SciDB array are closely analogous to primary key columns in a relational table. In an SQL database, you can relate two tables through a primary-key/foreign-key relationship. Analogously, you can relate two arrays along one or more shared dimensions.
One important difference: An index on an SQL table consumes space in the database. By contrast, SciDB indexing occurs automatically as a consequence of the array data model. That is, SciDB indexing requires no structures in the database beyond the arrays themselves.
That’s all for now. Stay tuned for more webinar follow-up.