vankesteren’s gists

vankesteren / metasynth.jl

Last active May 30, 2022 15:02

MetaSynth in Julia

	using Distributions
	using Random
	using DataFrames

	struct MetaVariable
	name::String
	p_missing::Float64
	dist::UnivariateDistribution
	end

vankesteren / decay_ar.R

Created April 1, 2022 07:11

exponential decay with AR(1)

	# exponential decay estimation with AR model
	N <- 30
	y <- numeric(N)
	first_obs <- 2 # where do we start?
	asymptote <- .5 # where do we go?
	ar1_param <- .6 # how slow do we go there? (between -1 and 1)
	sigma <- 0

	y[1] <- first_obs
	for (n in 2:N) {

vankesteren / glasso_emp_copula.R

Last active March 18, 2022 09:58

Penalized Gaussian copula with empirical marginals for data synthetization in R

	# Graphical LASSO copula model
	library(glasso)
	library(MASS)
	library(Matrix)
	set.seed(45)

	glasso_copula <- function(df, rho, penalize.diagonal = FALSE, ...) {
	N <- nrow(df)
	P <- ncol(df)
	if (missing(rho)) rho <- 2*sqrt(log(P)/N) # sls ch 9 p 252

vankesteren / vpglm.R

Last active February 23, 2022 13:25

fast vertically partitioned glms

vankesteren / rotation_translation.R

Last active December 16, 2021 14:58

Finding rotation (R) and translation (t) to transform a reference point set A into a target point set B = R * (A + t)

	set.seed(45)
	N <- 100
	P <- 2
	A <- matrix(rnorm(N*P), N)
	translation <- c(-0.9, .6)
	rotation <- function(theta) matrix(c(cos(theta), sin(theta), -sin(theta), cos(theta)), 2)
	R <- rotation(pi/5)
	B <- t(apply(A, 1, function(x) R %*% (x + translation)))

vankesteren / utr_house.R

Last active November 22, 2021 12:51

Where do people living in Utrecht look for a house?

	# load packages
	library(tidyverse)
	library(sf)

	# read data
	flow <- read_csv("https://data.mendeley.com/public-files/datasets/rtn8t47t6j/files/7f85a66d-a80d-4e3d-ada8-694a9c6e0a28/file_downloaded")
	gem <- st_read("WFS:https://geodata.nationaalgeoregister.nl/wijkenbuurten2018/wfs?&request=GetCapabilities&service=WFS",
	"wijkenbuurten2018:gemeenten2018")

	# clean gemeente dataset

vankesteren / tensorsem_lavpredict.R

Last active November 8, 2021 19:01

creating a lavaan object for using in lavPredict from a tensorsem parameter table

	# lavPredict from tensorsem parameters
	library(lavaan)
	library(tensorsem)

	mod <- "
	# three-factor model
	visual =~ x1 + x2 + x3
	textual =~ x4 + x5 + x6
	speed =~ x7 + x8 + x9
	"

vankesteren / em_mixture.R

Last active October 25, 2022 13:10

Unidimensional gaussian mixture modeling using EM

	# gaussian mixture modeling with EM
	priorp <- 0.6
	m1 <- 0
	m2 <- 2
	s1 <- 1
	s2 <- 0.707

	# generate some data with 2 classes
	N <- 1000
	cl <- rbinom(N, 1, priorp)

vankesteren / kmedoids.R

Created November 3, 2021 13:58

	kmedoids <- function(X, K, max_iter = 100) {
	X <- as.matrix(X)
	N <- nrow(X)
	P <- ncol(X)
	C <- matrix(0, nrow = K, ncol = P)
	clus <- sample(K, N, replace = TRUE)
	converged <- FALSE
	iter <- 0
	while (!converged && iter < max_iter) {
	old_clus <- clus

vankesteren / intransitivity_mc.R

Last active May 7, 2021 08:29

Bayesian monte carlo test for partial intransitivity

	# Bayesian test for partial intransitivity using monte-carlo posterior approximation
	# (c) Erik-Jan van Kesteren, 2021
	# Based on an e-mail discussion with Tony Marley
	library(expm)
	library(tidyverse)

	# Beta prior (1, 1 means uniform)
	prior_a <- 1
	prior_b <- 1

Erik-Jan van Kesteren vankesteren