learning journal
This page is a collection of small things I’ve learned over time. I find it useful to peruse to reinforce learning.
- JSON uses double-quotes.
- Type “iter
" to insert a for loop in IntelliJ.
- Java auto un boxing between primitive type and object type. For example:
double
toDouble
- EAV: entity attribute value (not just east Atlanta village)
- used jvisualvm to connect to a running java app, make a heapdump, and then peruse it locally
- check if rows 1 and 2 are exactly the same in excel:
=AND(EXACT(1:1, 2:2))
git log --grep
- you can edit
/etc/hosts
to map127.0.0.1
to something other than localhost
- frequent itemsets (e.g. buying diapers & beer at the same time)
- aka association rules
- “spoonerism”: the transposition of initial or other sounds of words, as in a blushing crow for a crushing blow.
- ICA (Independent Component Analysis):
- does not assume that components will be orthogonal, but relies on assumption of statistical independence. accordingly, it’s good for BSS (Blind Source Separation) — aka identifying the sources of multiple microphones.
- scikit-learn has FastICA
- vs. PCA:
- PCA wants each component to be gaussian, ICA wants each component to be as far from gaussian as possible
- PCA ranks components by how much each accounts for variance (so it can be used for dimensionality reduction), but ICA does not
- see: sci-kit learn and pg. 561 of Hastie, Tibshirani, Friedman
- how could i use this to get the filenames as output?
ls -lat *venues*.csv | awk '{print $10}' | xargs head -n 1 {} | grep "state"
- PS: use a for loop
- when printing with css media queries, use
position: absolute
instead offixed
; - scikit-learn doesn’t have a 1-sided 2-sample kolmogorov-smirnov statistic. why?
- Kleene Star:
/*
or([0-9]*)
(regex for “0” or “98765” or “”) - BNF: describe structure formally of, say, a postal e-mail address
strace
: debug kernel-level activities (the zine from julia — jvns.ca)- use
#
in css to target media queries for accessing the screen element to determine display style