19:10: We went to Nicholas. Decent place. Had a nice lamb kabob.
13:20: Parallel data analysis.
Wow. An expansive survey of the SOTA in Python parallelizing technologies: Concurrent.futures, Spark, Dask, etc. This is what a good tutorial does — surveys a taxonomy and illuminates the tradeoffs and overlaps. It can beat reading N articles by Y authors who are trying to convey Z messages. For some topics I don’t need this but for others I do.
Some of these technologies/tools are directly applicable to the data munging we do at Coffee Meets Bagel to fix database issues and migrate data for new features. Lots of things clicked into place!
My head filled up after about two hours. So many great technologies available, a fertile ground for innovation. Lots of parallel fun.
11:45: The morning tutorial was IPython and Jupyter in Depth.
I’ve never been very interested in either of these tools.
Re: Jupyter, I don’t often do interactive presentations, and when I do I just use a terminal window. I don’t often do plotting or interactive charting. I don’t write technical interactive papers. So while I appreciate what Jupyter provides, I don’t need it. Re: IPython, the only features that spark my interest are the multi-session command history and module reloading. I admire all the other features and marvel at their ingenuity, but I don’t feel a big draw.
I attended this tutorial because I wanted to ensure I wasn’t overlooking something. The tutorial was well done, and no, I’m not. I now know about more of their features and I get the appeal of the immersive environment for anyone doing frequent technical interactive presentations, writing technical papers, sharing data science, etc.
For my work, the benefits I’d get from IPython don’t outweigh the costs of keeping current with yet another tool. I guess I prefer using a smaller tool set that I know very well, than a larger tool set that I don’t know as well.
I can see where it’d be better deploy tutorial handouts as Jupyter notebooks rather than pdf files or stapled paper.
04:25: An almost-live post of the day.