rasmusab

# Test of Significance, takes the same arguments as t.test() .
signif.test <- function(x, ...) {
  p <- t.test(x, ...)$p.value
  
  # List of p excuses retrieved from http://mchankins.wordpress.com/2013/04/21/still-not-significant-2/
  p_excuses <- c(
    "(barely) not statistically significant <p>",
    "a barely detectable statistically significant difference <p>",
    "a borderline significant trend <p>",
    "a certain trend toward significance <p>",

rasmusab

Inlämningsuppgifter SM 1 & 2
========================================================
Rasmus Bååth, Sumsar Htååb

Detta är ett exempeldokumen skrivet med R markdown. 
R markdown-dokumentet måste vara fristående från din nuvarande R session. 
T.ex. så måste du i R markdown dokumentet alltid ladda in de paket du vill 
använda även om de redan är inladdade i din nuvarande R-session:

```{r message=FALSE}

rasmusab

library(rjags)
library(mvtnorm) # to generate correlated data with rmvnorm.
library(car) # To plot the estimated bivariate normal distribution.
set.seed(31415)

mu <- c(10, 30)
sigma <- c(20, 40)
rho <- -0.7
cov_mat <- rbind(c(     sigma[1]^2       , sigma[1]*sigma[2]*rho ),
                 c( sigma[1]*sigma[2]*rho,      sigma[2]^2       ))

rasmusab

# The code below will probably only work on Linux and requires VLC on the path.
library(beepr)
library(stringr)

# Plays a file with vlc and returns the vlc instance's PID
play_vlc_pid <- function(fname) {
  system(paste("vlc -Idummy --no-loop --no-repeat --playlist-autostart --no-media-library --play-and-exit", fname), 
         ignore.stdout = TRUE, ignore.stderr=TRUE,wait = FALSE)
  ps_out <- system("ps -eo pid,comm,etime| grep vlc", intern=TRUE)
  

rasmusab

# Prior to the tutorial make sure that the script below runs without error on your R installation.
# What you need is a working installation of JAGS: http://mcmc-jags.sourceforge.net/
# and the rjags R package that can be installed from R by running
# install.packages("rjags")

# Generating some fake data
set.seed(123)
y <- rbinom(30, size = 1, prob = 0.2015)

# Fitting a simple binomial model using JAGS

rasmusab

### Function for generating the posterior of Robo's position ###

landscape <- c(rep("plain", 50), rep("mountain", 25), rep("forest", 45))
landscape_color <- c(mountain = "black", forest = "green", plain = "yellow")
cover_cost <- c(mountain = 10, forest = 5, plain = 1)


posterior_sample <- function(n) {
  dist_i <- sample(3, n, replace = TRUE, c(18, 17, 65))
  mu <- c(15, 40, 90)

rasmusab

library(microbenchmark)
library(arm)
library(rstan)
library(bbmle)

log_post <- function(par, y, x) {
  sigma <- exp(par[1])
  intercept <- par[2]
  beta <- as.matrix(par[-c(1,2)])
  mu <- intercept + x %*% beta

rasmusab

# Converts the millionbase chess PGN database (http://www.top-5000.nl/pgn.htm) to json
# with one json dictionary per row. (That is, the resulting file is contain multiple json objects, 
# not just one large).

import json
import chess.pgn # From python-chess https://github.com/niklasf/python-chess

pgn = open("millionbase-2.22.pgn") # Or where you have put it
fout = open('milionbase-2.22.json', 'w') # Or where you want it

rasmusab

# Takes a json file describing chess games and produces a matrix with one row
# per turn showing how many pieces are left, one column per piece.

### Don't run this in R studio because it will take up twice the RAM 
### as R will make copies instead of references.

library(jsonlite)
library(stringi)

# Path to your json file as produced by this script: https://gist.github.com/rasmusab/07f1823cb4bd0bc7352d

rasmusab

# Takes the matrix created by this script: https://gist.github.com/rasmusab/fb98cced046d4c675d74
# and calculates some statistics and fits some models to it. All this is pretty memory heavy so saving
# the interesting parts using saveRDS so the script only need to be run once.

set.seed(123)
games <- as.data.frame(readRDS("milionbase_matrix.rds"))

# Saving some info
n_games <- max(games[,"game_id"])
fullmoves_white <- games[ games[, "fullmoves"] < 100 & games[,"active_player"] == 1, "fullmoves"]