randyzwitch

> #Inspect most popular words, minimum frequency of 20 
> findFreqTerms(dtm, lowfreq=20)
  [1] "15"           "2008"         "2009"         "2011"         "a"            "ad"           "add"          "adsens"      
  [9] "air"          "analyt"       "and"          "appl"         "at"           "back"         "bezel"        "black"       
 [17] "book"         "bookmark"     "break"        "broke"        "broken"       "bubbl"        "by"           "can"         
 [25] "case"         "chang"        "child"        "code"         "comment"      "comput"       "cost"         "cover"       
 [33] "crack"        "css"          "custom"       "data"         "delet"        "disabl"       "display"      "do"          
 [41] "doe"          "drop"         "edit"         "eleven"       "em209"        "entri"        "fix"          "footer"      
 [49] "footerphp"    "for"          "free"         "from"         "get"          "glue"         "googl"        "hadoop"      
 [57] "header"       "hing"         "how"          "i

randyzwitch

#I think there are 5 main topics: Data Science, Web Analytics, R, Julia, Wordpress
kmeans5<- kmeans(dtm, 5)

#Merge cluster assignment back to keywords
kw_with_cluster <- as.data.frame(cbind(searchkeywords$'Natural Search Keyword', kmeans5$cluster))
names(kw_with_cluster) <- c("keyword", "kmeans5")

#Make df for each cluster result, quickly "eyeball" results
cluster1 <- subset(kw_with_cluster, subset=kmeans5 == 1)
cluster2 <- subset(kw_with_cluster, subset=kmeans5 == 2)

randyzwitch

#Capitalize reserved keywords, indent between keywords and for multi-line statements
#Tables on separate lines
#Untested, hopefully this is valid SQL :)


SELECT
  a.key,
  a.col2,
  a.col3,
  COALESCE(a.col4,0) AS col4,

randyzwitch

#accumulator for cost results
cost_df <- data.frame()

#run kmeans for all clusters up to 100
for(i in 1:100){
  #Run kmeans for each level of i, allowing up to 100 iterations for convergence
  kmeans<- kmeans(x=dtm, centers=i, iter.max=100)
  
  #Combine cluster number and cost together, write to df
  cost_df<- rbind(cost_df, cbind(i, kmeans$tot.withinss))

randyzwitch

#Calculate lm's for emphasis
lm(cost_df$cost[1:10] ~ cost_df$cluster[1:10])
lm(cost_df$cost[10:19] ~ cost_df$cluster[10:19])
lm(cost_df$cost[20:100] ~ cost_df$cluster[20:100])

cost_df$fitted <- ifelse(cost_df$cluster <10, (19019.9 - 550.9*cost_df$cluster), 
                         ifelse(cost_df$cluster <20, (15251.5 - 116.5*cost_df$cluster),
                         (13246.1 - 35.9*cost_df$cluster)))

#Cost plot

randyzwitch

#Unzip file
unzip(zipfile= "/Volumes/32SDCARD/Data Sciences - General/SampleDailyDataFeed.zip", exdir= "~/Desktop/datafeed")

#Read in hit data
hit_data <- read.delim("~/Desktop/datafeed/hit_data.tsv", header=FALSE, stringsAsFactors = FALSE)

#Read in header to set column names
column_headers <- read.delim("~/Desktop/datafeed/column_headers.tsv")

#Set column headers for hit data to the column headers for 'column headers'

randyzwitch

library("RPostgreSQL")

#Connect to Redshift
redshift.Connection <- dbConnect(dbDriver("PostgreSQL"), host="XXXXXXXXXXXXXXX.redshift.amazonaws.com", dbname="XXXXXXXX",
                                 user="XXXXXXX", password= "XXXXXXXXXX", port=XXXXXXXX)

query_string <- 
"select *
from table
where something = 'True';"

randyzwitch

# Install the Redshift R library:
# https://github.com/pingles/redshift-r
# install.packages("~/Downloads/redshift-r-master", dependencies = T, repos = NULL, type = "source")

library(redshift)

redshift <- redshift.connect("jdbc:postgresql://REDSHIFT_DB:5439/DB_NAME", "LOGIN", "PASSWORD")

# Example Query:
data <- dbGetQuery(redshift, "SELECT COUNT(*) FROM table")

randyzwitch

#### Start IPython, generate SHA1 password to use for IPython Notebook server

$ ipython
Python 2.7.5 |Anaconda 1.8.0 (x86_64)| (default, Oct 24 2013, 07:02:20) 
Type "copyright", "credits" or "license" for more information.

IPython 1.1.0 -- An enhanced Interactive Python.
?         -> Introduction and overview of IPython's features.
%quickref -> Quick reference.
help      -> Python's own help system.

randyzwitch

library("RSiteCatalyst")
library("WriteXLS")

#Validate that underlying Perl modules for WriteXLS are installed correctly
#Will return "Perl found. All required Perl modules were found" if installed correctly
testPerl()

#### 1. Pull data for all report suites to create one comprehensive report ####

#Authenticate with Adobe Analytics API