4 ms·
High-Performance Click Analysis with MySQL
- emmett 18y agoI've had to independently discover many of these observations by hand while doing jtv stats as we've scaled...this is an excellent overview I wish I'd read 6 months ago.
- endtwist 18y agoI have a lot of experience with considering the tradeoffs necessary to do click analysis on a decently large scale (and the results of that will be shown, provided I can finish VisTrac). Analysis and collection presents a very interesting problem, given the mass of data being collected. Determining how much depth will be provided and how/when to aggregate is really important. I'm not advocating premature optimization, but this stuff requires a lot of structural consideration. The biggest hangup I've come across is filtering data by things like location/browser, and similar, as they would require aggregating data in many permutations...at least, as far as I've found. In the long run though, it seems that those levels of granularity aren't as important as they might seem and can often be dropped (as it will be, to an extent, in my case). (Random side-note : I'm potentially looking for a second brain to work on VisTrac with, as I haven't found one yet.)