comments encouraged..: September 2014

this week i

cobbled together a new method to noisify the replicate weights of a previously-linearized survey data set.

work with people who influence the editorial writers at the new york times.

consider the way the sequel to ripple. rolling stone magazine picked like a rolling stone as the greatest song of all time, nepotism at its worst.

see a zebra if this building were a rorschach test.

forgot some of those had those. and yes my camera has a rapid shutter speed, that isn't in free-fall.

sometimes think scientists ought to stop working on sex robots until the big stuff like hunger and wars have been eliminated. but then, on the other hand, maybe perhaps it's possible that the only way to snuff out global conflict is with an army of sex robots. so at ease, i guess. as you were.

left the city limits..

..for another bike ride through the woods.

circled the northern lake.

spotted a frog.

broke another one. i have snapped 75% of the chains on my used bicycle purchases in the past year.

put things in perspective.

9/25

this week i

saw the 2014 employer health benefits survey go live. matthew and nirmita do everything, i just watch.

lectured about survey analysis with the r language at the lithuanian institute of human geography and demography. then a few of us stuck around to eat cookies and drink coffee until i found myself with a mouthful of grinds dammit. also added anciasnapis to my quiver of platypi metaphors.

woke at 5:45 for a thirty-minute zombie-stare at this screen, boiled some broccoli, watched coursera lectures, exercised for an hour. that's monday, tuesday, thursday, friday. other mornings more nebulous. oh and i bought berlin-roasted beans from eastern baltic hipsters. but who gives a shit. here are the programming lecture notes:

x <- rnorm( 100 ) ; hist( x ) ; rug( x )
plot( 1 , 1 , pch = 20 ) ; legend( "topleft" , legend = "some text" , pch = 20 )
plot( 1 , 1 , type = "n" ) # plot everything except for the data
library(lattice) ; xyplot( Ozone ~ Wind | Month , data = airquality , layout = c( 5 , 1 ) )
names( x ) <- make.names( names( x ) ) # overwrite invalid column names with valid ones

reminded myself about extended bracketing. in the 2nd and 4th columns, blank out all values below zero.
    a <- sapply( rep( 10 , 4 ) , rnorm )
    a[ , c( 2 , 4 ) ][ a[ , c( 2 , 4 ) ] < 0 ] <- NA
    print( a )

shared a strawberry milkshake with a belarusian woman and her infinity shoelaces..

..we walked all of this, she didn't trip once. now you know as much as i do.

9/18

this week i

rolled out code to analyze the world values survey, famous for its gross national happiness metrics.

re-read almost every important number we've computed in the past year.

learned to uncheck the annotations box on youtube.

want to live somewhere that drivers smoke in their bus when they're off-duty. any place like that is okay by me.

saw a perfect duck. nothing else to report, just a perfect duck.

biked past the freshwater mermaid city limits (full route)..

fuel, service, alien abductions

pro(f/v)anity plates

..midday sky porn.

repeated the age-old words of the memoryless goldfish: wow, a castle.

9/11

this week i

worked my way through some data science specialization lectures. learned so far:

`grep( pattern , replacement , value = TRUE )` returns the elements rather than the positions
`mtcars[ , 1 , drop = FALSE ]` preserves the structure of the original object
lists can support non-exact match `x <- exact="FALSE]]`<br" h="" hi="1:5" list="" x="">
the dput() function has an optional output file parameter `dput( x , file = "out.R" )`

made public my preferred solution of long-to-wide reshaping.

preliminarily observed.. something in the cheese section.