Sam Swift swiftsam

VP Data Science @sealedinc previously led data @ArcadiaPower, @BoweryFarming, @Betterment and @goodjudgment | PhD in Organizational Behavior at Carnegie Mel

swiftsam / cac_fig4.R

Last active December 1, 2017 04:27

ggplot code to reproduce figure 4 from https://doi.org/10.1287/mnsc.2016.2525

	library(ggplot2)
	library(data.table)
	library(scales)


	ct.binned <- overp.indiv.ct[, list(confidence = mean(MeanMaxFcast),
	hit_rate = mean(HitRate),
	ci_hit_rate = 1.96*sd(HitRate)/sqrt(.N),
	count = .N),
	by=c("bin,ct")]

swiftsam / r_cache.R

Created May 6, 2016 15:57

function caching function for R

	####~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
	### Cache (memoization) utilities
	###
	### Purpose
	### * Functions to save the output of time consuming functions to memory and
	### return results quickly if we already have the answers
	###
	### Notes: Standard usage would go like this
	### FunctionName <- function(){
	### args <- c(as.list(environment()), list())

swiftsam / remove_dup__packages.R

Created February 9, 2016 16:47

remove R packages installed in multiple libraries

	user.libs <- row.names(installed.packages(lib.loc = .libPaths()[1]))
	sys.libs <- row.names(installed.packages(lib.loc = .libPaths()[2]))

	remove.packages(intersect(sys.libs,user.libs), lib=.libPaths()[2])

swiftsam / ordinal_brier.R

Created August 4, 2014 16:11

ordinal brier scoring

	###~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
	### calculates Brier Score for a single forecast on an Ordered IFP
	### calculation based on definition in Appendix B of ACE_TE_Plan_OY1_v3.
	### procedure reduces penalty to probability assigned to bins adjacent
	### to the realized outcome
	###~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
	ScoreOrdered <- function(fcast.values,
	correct.values,
	scoring.fn){
	# Step 1 – Take the original categories (A-B-C-D) and break them up