andland

roll_die <- function(status, strategy) {
    roll = sample(c(names(status), "basket"), 1)
    if (roll == "basket") {
        trees = status[names(status) != "raven" & status > 0]
        if (strategy == "optimal") {
            biggest_trees = trees[trees == max(trees)]
            roll = sample(names(biggest_trees), 1)
        } else if (strategy == "random") {
            roll = sample(names(trees), 1)
        } else if (strategy == "worst") {

andland

Below are links to manuscripts or packages related to my talk at JSM on Supervised Dimensionality Reduction. Session info here.

My slides will be here.

Contact:

• My Website
• Twitter
• Github
• Yoonkyung Lee's Website

andland

I will not share anyone's information. I will only use aggregated data with no personal information.

andland

Below are links to manuscripts or packages related to my talk at JSM on Generalized PCA. Session info here.

My slides are here.

• Logistic PCA manuscript
  • R package logisticPCA
• Generalized PCA manuscript
  • R package generalizedPCA

andland

andland

See this link for an introduction on time stacking and time slicing.

time_slice.R requires the number of pixels wide or tall the image is to be a multiple of the number of images in your timelapse.

time_slice_v2.R attempts to get around this. Some images will contribute more pixels per slice than others. This is done by making the first x% of the images cover the first x% of the pixels (with appropriate rounding). It does not deal with number of images being greater than the height or width of the images in pixels. Version 2 will probably work better for you.

For example, if the images are 150 pixels wide and your timelapse has 100 images, time_slice.R will make the first image have a slice which is 51 pixels wide. The remaining 99 images will get slices which are 1 pixel wide. time_slice_v2.R will alternate between 1 pixel per i

andland

library(XML)
library(lubridate)
library(sqldf)
library(reshape2)
library(ggplot2)
library(mgcv)

cat("loading old data...\n")
playlist=read.csv("CD101Playlist.csv",stringsAsFactors=FALSE)
colnames(playlist)[3]="Last Played"

andland

library(shiny)
library(ggplot2)
exp.age.df=read.csv("https://dl.dropboxusercontent.com/u/17648661/ExpAgeByNameYear.csv")

age.range=range(exp.age.df$Age)

unique.names=sort(unique(exp.age.df$Name))
unique.names=c("<NONE>",as.character(unique.names))

start.names=c("Andrew","Dylan","Fred","Grace","Lillian","John")

andland

# rm(list=ls())
setwd("Kaggle/Amazon Employee")

train = read.csv("train.csv")
test = read.csv("test.csv")
train$ROLE_TITLE <- NULL # Because the same as ROLE_CODE
test$ROLE_TITLE <- NULL # Because the same as ROLE_CODE

jaccard <- function(vec, matrix) {
  rowSums(as.matrix(sweep(matrix, 2, as.numeric(vec), "==")))

andland

# inspired by http://schamberlain.github.io/2012/01/logistic-regression-barplot-fig/
logithistplot <- function(data,breaks="Sturges",se=TRUE) {
    require(ggplot2);
    col_names=names(data)
    
    # get min and max axis values
    min_x <- min(data[,1])
    max_x <- max(data[,1])
    
    # get bin numbers