UFO Sightings Analysis with R

"R is a free software programming language and software environment for statistical computing and graphics." 

-Wikipedia

• Dialect of S language
• Free software
• In 2000, R version 1.0.0 is released

• Install RStudio
• Using common features of RStudio
• Load a csv table
• Use zoo package for dates data type
• Subset data
• Draw insights from data!

1. Getting RStudio: http://www.rstudio.com/products/rstudio/download/

2. Customizing your GUI

3. Alternative: RGui from http://www.r-project.org/

Hotkeys:

• up and down for last or recent line of code
• Ctrl + r is for running a line of code
• Ctrl + l is for clearing the console

• install packages
• RStudio has help tab
• set working directory
• preview data
• executing R code - try 2+3
• save and open your R file
• save data

A very hotkey stroke:

• “tab” for predictive keyboard

There are several methods:

1) use RStudio GUI to set working directory

2) use setwd()

3) your csv file name in read.csv() contains the entire directory

nasdaqData <- read.csv("~/Github/gdi-r/class1/datasets/NASDAQOMX-NDX.csv")

There are bunch of arguments (ie. options)! Here is my advice for best practice:

setwd("~/Github/gdi-r/class1/datasets")
opecData <- read.csv("OPEC-ORB.csv", header = T, sep = ",", stringsAsFactors = F) #loading csv file

opecData <- read.csv("OPEC-ORB.txt", header = T, sep = "\t", stringsAsFactors = F) #loading txt file

PS. Data sets were taken from quandl.com

2 ways to find out what kind of table you have

(matrix, list, data frame, etc.)

Method 1:

str(nasdaqData)

Method 2: Go to Environment tab

• Method 1: using subset()
• Method 2: using square brackets (ie. extraction brakets)

let's say we only want to see part of the data with index value higher than 4000

highIndexValue <- subset(nasdaqData, Index.Value > 4000)

highIndexValueBrackets <- nasdaqData[nasdaqData$Index.Value > 4000 , ]

Comma is needed because brackets have 2 arguments: the row and the column

[row, col]

The criteria that we're using tells R which row to extract!

Your criteria goes after the comma!

justACol <- nasdaqData[, "Low"]

You can use the column name with quotes or the number of the column.

Now you have a vector, justACol, that contains everything that is in Low column from nasdaqData

use summary() to get summary statistics

summary(nasdaqData)
summary(highIndexValue)

http://blog.modeanalytics.com/five-public-dataset/

• Which month is most popular for UFO sighting?

ufoData$month <- as.Date(ufoData$month, format = "%Y-%m-%d")

use summary() to get summary statistics

summary(ufoData)

There are many ways to solve this problem.

Plan:

• Aggregate sightings by month
• Combind all the months together to form a data frame (ie. table)
• Use summary() to find the mean and median sightings for each month

OR

• We can use summary() for each month, then combind the summaries

Problem with the first plan is that different months, we have diffent amount of sightings.

Can't use cbind with vectors that have different lenths.

There are ways around this by inputting “NA”, but let's just use the second method…

Use what we know about:

• summary()
• subset()
• square brackets

Put them all together!

Hint: format.Date(month, “%m”) == “01” is the criteria that we have to use to tell subset function to select rows with month = 01.

Jan <- summary(subset(ufoData, format.Date(month, "%m") == "01" )[,"sightings"])
Feb <- summary(subset(ufoData, format.Date(month, "%m") == "02" )[,"sightings"])
Mar <- summary(subset(ufoData, format.Date(month, "%m") == "03" )[,"sightings"])
Apr <- summary(subset(ufoData, format.Date(month, "%m") == "04" )[,"sightings"])

… replicate each line for each month of the year

We can cover how to write functions in a different R course

summMonth <- function(mm){
  summary(subset(ufoData, format.Date(month, "%m") == mm )[,"sightings"])
}

We have to input month name and corresponding month number…

Jan <- summMonth("01")
Feb <- summMonth("02")
Mar <- summMonth("03")
Apr <- summMonth("04")
May <- summMonth("05")
Jun <- summMonth("06")
Jul <- summMonth("07")
Aug <- summMonth("08")
Sep <- summMonth("09")
Oct <- summMonth("10")
Nov <- summMonth("11")
Dec <- summMonth("12")

ufoDataByMonth <- cbind(Jan, Feb, Mar, Apr, May, Jun, Jul, Aug, Sep, Oct, Nov, Dec)

Check out ufoDataByMonth table!

colnames(ufoDataByMonth)[ufoDataByMonth["Mean",] == maxMean]

[1] "Jul"

maxMedian <- max(ufoDataByMonth["Median",])
colnames(ufoDataByMonth)[ufoDataByMonth["Median",] == maxMedian]

[1] "Jun"

1) A foundation course

Understanding data types for data cleansing

2) Best of R

Intro to statistics and machine learning with R

3) Dataviz with R

Learn how to visualize your data with the following R packages: ggplot2 & shiny

We will go through this tutorial together: http://shiny.rstudio.com/tutorial/lesson5/

or we can do something else!

I'm open to suggestions for the type of R class you're interested! Also, I'm open to using any types of data set.