``` timestamp=1 idx=2 key=foo key2=bar key3=4 rank=2 timestamp=2 idx=3 key=baz key2=box key3=4 rank=3 timestamp=3 idx=3 key=baz key2=box key3=4 rank=3 mode=start timestamp=4 idx=3 key=baz key2=box key3=4 rank=3 mode=end ```
and I'd like to do a bunch of adhoc / custom graphing. For example: there's the obvious stuff like graphing how many times I had the value "baz" for "key2" in "rank=3" per minute.
But there's also less obvious stuff like "graph the average time between mode=start/mode=end" and calculate the average delay between an "mode=end" and the next "mode=start" per rank.
So far I've been doing pretty well with a jupyter notebook + pandas, but it's pretty verbose, and I don't think matplotlib is the best graphing software. Also, pandas sucks.
Are there better solutions for graphing? I was looking into gnuplot, but its API seems unwieldy.
I'm looking for something lightweight (my logs are only on the order of ~ 50 - 500 MB). Not trying to setup Prometheus / Loki, or anything of similar complexity.
Not easy to make recommendations because lot of missing context -- diy / personal improvement, work related, eduational related, need something for a presentation discussion at monthly neighborhood meet up of retired NASA statisicians trying to learn non-mainframe / non-proprietary / newer than voyager I 45 year old data format ways of parsing log data & don't have access to modern pc equipment.
Note: make sure not violating terms of usage if go the route of find demos/examples of things similar what trying to generate, modify to situation as needed.
old school, light weight, but need some backround in programming/scripting: gawk/shell commands in a script & generate/feed to nroff/graph. more modern: latex/gnuplot/ggplot2/gnu octive/R
IMHO: Things only seems verbose if have never used before & trying to use non-default settings.