IronistM

ip2Int <- function(ip){
  split <- as.numeric(strsplit(ip, "\\.")[[1]])
  out <- (split[1] * 256^3) + (split[2] * 256^2) + (split[3] * 256) + (split[4])
  return(out)
}


int2Ip <- function(int){
  split <- NULL
  split[1] <- as.integer(int/256^3)

IronistM

#Plot data using ggplot2
library(ggplot2)

#Combine year/month/day together into POSIX
pageviews_w_forecast$date <- ISOdate(pageviews_w_forecast$year, pageviews_w_forecast$month, pageviews_w_forecast$day)

#Convert columns to numeric
pageviews_w_forecast$pageviews <- as.numeric(pageviews_w_forecast$pageviews)
pageviews_w_forecast$pageviews_upper <- as.numeric(pageviews_w_forecast$pageviews_upper)
pageviews_w_forecast$pageviews_lower <- as.numeric(pageviews_w_forecast$pageviews_lower)

IronistM

LOAD
if(age<16, 'Under 16',
     if(age>65, 'Over 65',
          text(16+(Div(age-16,10)*10)) & ' - ' & text(25+(Div(age-16,10)*10)))) as age_group

IronistM

// in this case we want to load a 10% sampling set of the data
data:
LOAD ….<your fields>…
FROM <your QVD file>
WHERE ceil(rand() * 100) <= 10;  // or: rand() <= 0.1

// SAMPLE n LOAD is much faster but allocates the memory of the full data set during the LOAD
// Where rand() <= n allocates only the memory of the smaller result set which could be crucial with Big Data..

IronistM

// Use Gists to store code you would like to remember later on
console.log(window); // log the "window" object to the console

IronistM

ANOVA<-function(fit1,fit2){
  temp <- anova(fit2,fit1 )
  fin.aov <- anova(fit1)   
  reg <- temp[2,2:6]
  rownames(reg) <- "Regression"
  reg[1,1:2] <- reg[1,2:3]
  reg[1,3] <- reg[1,2]/reg[1,1]
  colnames(reg) <- colnames(fin.aov)
  res <- fin.aov[tail(nrow(fin.aov),1),]
  tot <- cbind(reg[1,1:2]+res[1,1:2],NA,NA,NA)

IronistM

suppressMessages(library(forecast))

data<-read.csv( file('stdin') )

anomaly_detection<-function(data){
  
  seasonality<-48
  data_series<-ts(data$count,frequency=seasonality)
  
  train_start<-1 ## train on 1 month of data

IronistM

####################
# Create relogit predicted probabilities using Zelig and ggplot2
# Two Sword Lengths: Losers' Consent and Violence in National Legislatures (Working Paper 2012)
# Christopher Gandrud
# Updated 26 April 2012
###################

## Load required packages
library(RCurl)
library(Zelig)

IronistM

N <- 10
id  <-  1:10
x  <- 1 + rnorm(N) - 1*id
date  <- seq(as.Date("2013-07-01"), by = "year", along = x)
df  <- data.frame(x = x, date = date)
plot(df$date, df$x)
summary(lm(x ~ date, data = df))

IronistM

# Library Loading
library("RPostgreSQL");
library("car");

# Connect to Database
pgDrv <- dbDriver("PostgreSQL")
dbh <- dbConnect(pgDrv, host="localhost", dbname="dnsmonitor", user="dnsmon", password="tooEasy")

# Retrieve Statistics from DB
stats <- dbGetQuery(dbh, "select client.id, client.ip, sum(queries) as queries, sum(nx) as nx, sum(answers) as answers, sum(errors) as errors, count(distinct day) as days_active