CSSS 508, Week 10
Rebecca Ferrell
June 1, 2016
Topics
- Scraping the web with
rvest - Mining text with
tm - What next?
Web scraping with rvest
Wait, isn't that Argus Filch?
Game of Thrones x Harry Potter
We'll use the package rvest (“harvest”) to grab IMDb casts for Game of Thrones and Harry Potter to identify all overlapping actors.
# install.packages("rvest")
library(rvest)
First, try out SelectorGadget.
# pull full Game of Thrones cast page
got_page <- read_html("http://www.imdb.com/title/tt0944947/fullcredits")
got_cast_raw <- got_page %>%
html_nodes(".itemprop .itemprop , .character div") %>%
html_text()
Cleaning up the Game of Thrones cast
head(got_cast_raw)
[1] "Peter Dinklage"
[2] "\n  Tyrion Lannister\n (56 episodes, 2011-2016)\n "
[3] "Lena Headey"
[4] "\n  Cersei Lannister\n (50 episodes, 2011-2016)\n "
[5] "Kit Harington"
[6] "\n  Jon Snow\n (47 episodes, 2011-2016)\n "
Pattern appears to be: [actor name], [messy character info], repeat.
Cleaning up the Game of Thrones cast
Make a data frame:
got_cast_df <- data.frame(matrix(got_cast_raw, ncol = 2, byrow = TRUE), stringsAsFactors = FALSE)
colnames(got_cast_df) <- c("Actor", "char_info")
head(got_cast_df, 3)
Actor
1 Peter Dinklage
2 Lena Headey
3 Kit Harington
char_info
1 \n  Tyrion Lannister\n (56 episodes, 2011-2016)\n
2 \n  Cersei Lannister\n (50 episodes, 2011-2016)\n
3 \n  Jon Snow\n (47 episodes, 2011-2016)\n
Clean up the character column
We want to trim initial whitespace, extract the character, and move the episode count to a new column.
library(stringr); library(dplyr)
got_cast <- got_cast_df %>%
mutate(char_info = str_trim(char_info),
GoT_character = str_trim(str_extract(char_info, "^.*\\n")),
Episodes = as.numeric(str_extract(str_extract(char_info, "[0-9]* episode"), "[0-9]*"))) %>%
select(-char_info)
Scraping Harry Potter actors
We'll want to loop over all eight films to do this!
HP_URLs <- c("http://www.imdb.com/title/tt0241527/fullcredits", "http://www.imdb.com/title/tt0295297/fullcredits", "http://www.imdb.com/title/tt0304141/fullcredits", "http://www.imdb.com/title/tt0330373/fullcredits", "http://www.imdb.com/title/tt0373889/fullcredits", "http://www.imdb.com/title/tt0417741/fullcredits", "http://www.imdb.com/title/tt0926084/fullcredits", "http://www.imdb.com/title/tt1201607/fullcredits")
Harry Potter scraping
Looping game plan:
- Create a list with a spot for each film
- Scrape the cast into the spot for each film
- Reshape the character vector into a matrix
- Combine the casts of all the films
- Remove whitespace, etc.
HP_cast_list <- vector("list", length(HP_URLs))
Looping
Consolidate the work done for GoT into a loop for HP:
for(i in seq_along(HP_URLs)) {
HP_cast_list[[i]] <- read_html(HP_URLs[i]) %>%
html_nodes(".itemprop .itemprop , .character div") %>%
html_text() %>%
matrix(ncol = 2, byrow = TRUE) %>%
data.frame(stringsAsFactors = FALSE)
colnames(HP_cast_list[[i]]) <- c("Actor", "HP_character")
}
HP_cast <- bind_rows(HP_cast_list, .id = "HP_film") %>%
mutate_each(funs(str_trim))
Who was in both?
both_GoT_HP <- HP_cast %>%
inner_join(got_cast, by = "Actor") %>%
arrange(desc(Episodes), Actor)
Other ways of getting data off the web
Specialized packages for specific services:
twitteR(Twitter REST API),streamR(Twitter streaming API),Rfacebook- Require you get a key to run queries – store in separate file and pull in, do not hardcode/share with others!
- Rate limiting can be challenge, use
Sys.sleep(seconds)if needed to slow code down
General API access:
httrfor HTTP requests and responsesjsonlitefor parsing JSON,XMLfor XML
Many tutorials just a Google search away!
Text mining with tm
Text mining terminology
tm: R package for performing text mining- Term: word
- Document: collection of terms
- Corpus: a collection of documents (plural: corpora)
- Dictionary: set of relevant terms
My first corpus
We can make a toy corpus manually by creating a character vector, running VectorSource on it to read it in, and then VCorpus to corpus-ify:
library(tm)
UW_tweets <- c("Remembering and honoring those who made the ultimate sacrifice while serving our country. #MemorialDay2016", "VIDEO: This spring @UW students taught literacy arts to #Colville Reservation students. Check out book they made!", "Enjoy the long weekend, Huskies! And to those studying for finals: Good luck and hang in there!", ".@UWBuerk & @UWFosterSchool–hosted biz plan competition awards $85,000 to students for new ventures. http://ow.ly/3PtI300F87Y #UWinnovates")
toy_corpus <- VCorpus(VectorSource(UW_tweets))
Accessing corpus entries
A corpus is just a fancy list of documents, and you can access a document as a list entry:
toy_corpus[[3]]
<<PlainTextDocument>>
Metadata: 7
Content: chars: 95
as.character(toy_corpus[[3]])
[1] "Enjoy the long weekend, Huskies! And to those studying for finals: Good luck and hang in there!"
Text files as documents
You will more likely be making corpora from sources like Twitter or reading in data from text files.
We'll import a sample of emails from the Enron corpus assembled by UC Berkeley students. First, let's download a ZIP file with the text files and unzip it.
download.file("https://www.dropbox.com/s/qrd1j44qnlzg68a/enron_sample_emails.zip?dl=1", destfile = "enron_emails.zip", mode = "wb")
unzip("enron_emails.zip", exdir = "enron_emails")
length(list.files("enron_emails/enron_sample_emails"))
[1] 100
Reading in text files
Make a corpus where each document is an email in the Enron subsample:
enron_corpus <- VCorpus(DirSource(directory = "enron_emails/enron_sample_emails", mode = "text"))
as.character(enron_corpus[[3]])
[1] "Message-ID: <26658543.1075843310470.JavaMail.evans@thyme>"
[2] "Date: Fri, 22 Dec 2000 04:00:00 -0800 (PST)"
[3] "From: sarah.novosel@enron.com"
[4] "To: susan.mara@enron.com"
[5] "Subject: Re: Wolak report"
[6] "Cc: donna.fulton@enron.com, jeff.dasovich@enron.com"
[7] "Mime-Version: 1.0"
[8] "Content-Type: text/plain; charset=us-ascii"
[9] "Content-Transfer-Encoding: 7bit"
[10] "Bcc: donna.fulton@enron.com, jeff.dasovich@enron.com"
[11] "X-From: Sarah Novosel"
[12] "X-To: Susan J Mara"
[13] "X-cc: Donna Fulton, Jeff Dasovich"
[14] "X-bcc: "
[15] "X-Folder: \\Jeff_Dasovich_June2001\\Notes Folders\\All documents"
[16] "X-Origin: DASOVICH-J"
[17] "X-FileName: jdasovic.nsf"
[18] ""
[19] "Sue:"
[20] ""
[21] "I'm so impressed that you know how to put the link on the email. I can't "
[22] "figure out how to do that, and I think that looks much more \"high tech\" than "
[23] "attaching a silly old report."
[24] ""
[25] "Thanks for the link. We can take it from here. "
[26] ""
[27] "Sarah"
[28] ""
[29] ""
[30] ""
[31] ""
[32] ""
[33] "\tSusan J Mara"
[34] "\t12/22/2000 11:31 AM"
[35] "\t\t "
[36] "\t\t To: Sarah Novosel/Corp/Enron@ENRON, Donna Fulton/Corp/Enron@ENRON"
[37] "\t\t cc: Jeff Dasovich/NA/Enron@Enron"
[38] "\t\t Subject: Wolak report"
[39] ""
[40] "Call me computer illiterate. I could not figure out how to download the "
[41] "report off the web site, so here's the link."
[42] ""
[43] " "
[44] " http://www.caiso.com/docs/2000/12/12/2000121215070918957.pdf"
Transformations (maps)
Let's change to lowercase, remove “stopwords” and header terms, remove punctuation, numbers, and whitespace, and “stem” the words:
# install.packages("SnowballC") # may solve errors
enron_stripped <- enron_corpus %>%
tm_map(content_transformer(str_to_lower)) %>%
tm_map(removeWords, stopwords("english")) %>%
tm_map(removeWords, c("javamail.evans@thyme", "message-id", "date", "subject", "mime-version", "content-type", "text/plain", "charset=us-ascii", "content-transfer-encoding", "x-", "x-cc", "x-bcc", "x-folder", "x-origin", "x-filename")) %>%
tm_map(removePunctuation) %>%
tm_map(removeNumbers) %>%
tm_map(stripWhitespace) %>%
tm_map(stemDocument)
Word clouds
library(wordcloud)
wordcloud(enron_stripped, min.freq = 2, max.words = 80)
Filtering to emails with California
We'll write a function that takes the content of the documents and looks for any instance of "california", then use it with tm_filter:
doc_word_search <- function(x, pattern) {
any(str_detect(content(x), pattern = pattern))
}
cali_emails <- enron_stripped %>%
tm_filter(doc_word_search, pattern = "california")
length(cali_emails)
[1] 14
Term-Document Matrices
We can look for patterns across the documents by constructing a TermDocumentMatrix:
enron_tdm <- TermDocumentMatrix(enron_stripped)
str(enron_tdm)
List of 6
$ i : int [1:17721] 2 9 12 15 17 19 24 27 30 34 ...
$ j : int [1:17721] 1 1 1 1 1 1 1 1 1 1 ...
$ v : num [1:17721] 4 1 1 1 1 2 1 1 2 4 ...
$ nrow : int 5061
$ ncol : int 100
$ dimnames:List of 2
..$ Terms: chr [1:5061] "abat" "abil" "abl" "abnorm" ...
..$ Docs : chr [1:100] "114495.txt" "115323.txt" "118675.txt" "119214.txt" ...
- attr(*, "class")= chr [1:2] "TermDocumentMatrix" "simple_triplet_matrix"
- attr(*, "weighting")= chr [1:2] "term frequency" "tf"
What does the matrix look like?
Too big to view at once, but we can look at snippets with inspect:
inspect(enron_tdm[1:5, 1:5])
<<TermDocumentMatrix (terms: 5, documents: 5)>>
Non-/sparse entries: 2/23
Sparsity : 92%
Maximal term length: 7
Weighting : term frequency (tf)
Docs
Terms 114495.txt 115323.txt 118675.txt 119214.txt 12030.txt
abat 0 0 0 0 0
abil 4 0 0 0 0
abl 0 0 0 0 1
abnorm 0 0 0 0 0
abolish 0 0 0 0 0
Removing sparse words
We could focus on words that appear in at least 40% of documents.
enron_tdm_sparse <- removeSparseTerms(enron_tdm, 0.60)
inspect(enron_tdm_sparse)
<<TermDocumentMatrix (terms: 11, documents: 100)>>
Non-/sparse entries: 659/441
Sparsity : 40%
Maximal term length: 18
Weighting : term frequency (tf)
Docs
Terms 114495.txt 115323.txt 118675.txt 119214.txt 12030.txt
bit 1 1 1 1 1
document 2 1 1 1 0
folder 1 1 1 1 0
forward 3 1 0 0 0
kean 0 2 0 0 0
pdt 1 1 0 0 1
skeannsf 0 0 0 0 0
steven 0 3 0 0 0
stevenkeanenroncom 0 1 0 0 0
thank 2 0 1 0 0
will 19 3 0 0 0
Docs
Terms 12174.txt 121748.txt 121756.txt 12176.txt 122913.txt
bit 1 1 1 2 1
document 0 1 1 0 1
folder 0 1 1 0 1
forward 0 0 0 0 0
kean 0 0 0 0 0
pdt 0 1 1 0 1
skeannsf 0 0 0 0 0
steven 0 0 0 0 0
stevenkeanenroncom 0 0 0 0 0
thank 1 0 0 0 0
will 0 0 0 7 0
Docs
Terms 122926.txt 130135.txt 149993.txt 165052.txt
bit 1 1 1 1
document 1 1 1 0
folder 1 1 1 0
forward 0 0 1 0
kean 0 0 0 3
pdt 1 0 0 0
skeannsf 0 0 0 0
steven 0 0 0 2
stevenkeanenroncom 0 0 0 1
thank 0 1 0 1
will 1 1 0 0
Docs
Terms 173408.txt 173412.txt 173765.txt 173769.txt
bit 1 1 1 1
document 1 1 1 1
folder 1 1 1 1
forward 0 1 1 0
kean 2 2 2 2
pdt 0 0 1 1
skeannsf 1 1 1 1
steven 2 3 2 2
stevenkeanenroncom 1 1 1 1
thank 1 0 0 1
will 0 0 2 1
Docs
Terms 173801.txt 173839.txt 173842.txt 173884.txt
bit 1 1 1 1
document 1 1 1 1
folder 1 1 1 1
forward 2 0 1 1
kean 2 2 2 2
pdt 1 1 1 1
skeannsf 1 1 1 1
steven 3 2 3 3
stevenkeanenroncom 1 1 1 1
thank 0 0 1 1
will 0 1 4 0
Docs
Terms 173904.txt 173955.txt 173995.txt 174075.txt
bit 1 1 1 1
document 1 1 1 1
folder 1 1 1 1
forward 0 1 1 1
kean 2 2 2 3
pdt 1 1 1 1
skeannsf 1 1 1 1
steven 2 1 2 4
stevenkeanenroncom 1 1 1 1
thank 0 3 1 0
will 1 0 0 1
Docs
Terms 174092.txt 174123.txt 174137.txt 174212.txt
bit 1 1 1 0
document 1 1 1 1
folder 1 1 1 1
forward 1 1 0 1
kean 2 2 2 2
pdt 1 1 1 1
skeannsf 1 1 1 1
steven 3 3 2 2
stevenkeanenroncom 1 1 1 1
thank 1 2 2 0
will 1 2 1 8
Docs
Terms 174376.txt 174394.txt 174517.txt 175076.txt
bit 1 1 1 1
document 1 1 1 1
folder 1 1 1 1
forward 1 1 5 0
kean 2 2 2 2
pdt 1 1 1 1
skeannsf 1 1 1 1
steven 2 3 3 2
stevenkeanenroncom 1 1 1 1
thank 1 1 1 1
will 2 0 6 0
Docs
Terms 175159.txt 175162.txt 175163.txt 175204.txt
bit 1 1 1 1
document 1 1 2 1
folder 1 1 1 1
forward 1 0 1 1
kean 2 2 2 3
pdt 1 1 1 1
skeannsf 1 1 1 1
steven 2 2 3 1
stevenkeanenroncom 1 1 1 1
thank 3 2 1 2
will 1 0 0 0
Docs
Terms 175209.txt 175217.txt 175222.txt 175240.txt
bit 1 1 1 1
document 1 1 1 1
folder 1 1 1 1
forward 2 0 2 1
kean 2 3 2 2
pdt 1 1 1 1
skeannsf 1 1 1 1
steven 3 4 3 2
stevenkeanenroncom 1 1 1 1
thank 0 0 1 1
will 0 1 0 0
Docs
Terms 175243.txt 175246.txt 175418.txt 175446.txt
bit 1 1 1 1
document 1 1 1 1
folder 1 1 1 1
forward 1 0 0 0
kean 2 2 2 2
pdt 1 1 1 1
skeannsf 1 1 1 1
steven 2 1 1 2
stevenkeanenroncom 1 1 1 1
thank 2 1 0 3
will 1 0 0 0
Docs
Terms 175493.txt 175501.txt 175596.txt 175607.txt
bit 1 1 1 1
document 1 1 1 3
folder 1 1 1 1
forward 2 0 1 8
kean 2 3 2 1
pdt 1 1 0 0
skeannsf 1 1 1 1
steven 3 4 3 0
stevenkeanenroncom 1 1 1 0
thank 1 0 1 1
will 1 0 0 73
Docs
Terms 175617.txt 175619.txt 175621.txt 175622.txt
bit 1 1 1 1
document 1 1 1 1
folder 1 1 1 1
forward 1 0 1 1
kean 2 3 2 3
pdt 0 0 0 0
skeannsf 1 1 1 1
steven 3 2 3 5
stevenkeanenroncom 1 1 1 1
thank 4 1 2 0
will 2 1 0 2
Docs
Terms 175662.txt 175810.txt 176524.txt 176539.txt
bit 1 1 0 1
document 1 1 1 1
folder 1 1 1 1
forward 0 1 2 1
kean 2 3 3 2
pdt 0 0 1 1
skeannsf 1 1 1 1
steven 2 4 1 3
stevenkeanenroncom 1 1 1 1
thank 1 0 1 0
will 1 0 1 0
Docs
Terms 176580.txt 176591.txt 176601.txt 176727.txt
bit 1 1 1 1
document 1 1 1 1
folder 1 1 1 1
forward 1 1 0 0
kean 2 2 2 2
pdt 1 1 1 1
skeannsf 1 1 1 1
steven 2 2 1 1
stevenkeanenroncom 1 1 1 1
thank 3 0 4 3
will 0 3 5 0
Docs
Terms 176805.txt 177215.txt 177800.txt 177819.txt
bit 1 1 1 1
document 2 2 0 0
folder 0 0 0 0
forward 8 8 0 1
kean 1 1 2 2
pdt 0 0 1 1
skeannsf 1 1 0 0
steven 0 0 3 4
stevenkeanenroncom 0 0 1 1
thank 1 1 0 0
will 73 73 1 0
Docs
Terms 177847.txt 177853.txt 178057.txt 194901.txt
bit 1 1 1 1
document 0 0 0 1
folder 0 0 0 0
forward 0 0 0 0
kean 2 2 2 0
pdt 1 1 1 1
skeannsf 0 0 0 0
steven 3 3 3 0
stevenkeanenroncom 1 1 1 0
thank 1 0 1 0
will 0 0 2 0
Docs
Terms 229326.txt 239959.txt 255511.txt 50307.txt 52201.txt
bit 1 1 2 1 1
document 0 0 2 0 0
folder 0 0 1 0 0
forward 2 0 3 0 0
kean 2 0 0 0 0
pdt 1 0 1 1 1
skeannsf 0 0 0 0 0
steven 1 0 0 1 0
stevenkeanenroncom 0 0 0 0 0
thank 1 2 2 0 0
will 0 0 19 0 0
Docs
Terms 52713.txt 53536.txt 54263.txt 54536.txt 54537.txt
bit 1 1 1 1 1
document 0 0 0 5 0
folder 0 0 0 0 0
forward 0 0 0 1 0
kean 0 0 0 0 0
pdt 0 0 1 1 1
skeannsf 0 0 0 0 0
steven 0 0 0 0 0
stevenkeanenroncom 0 0 0 0 0
thank 1 1 1 3 1
will 0 0 0 3 1
Docs
Terms 54540.txt 54566.txt 54577.txt 54582.txt 54595.txt
bit 1 1 1 1 1
document 0 0 0 0 0
folder 0 0 0 0 0
forward 0 0 1 0 1
kean 0 0 0 0 0
pdt 1 1 1 1 1
skeannsf 0 0 0 0 0
steven 0 0 0 0 0
stevenkeanenroncom 0 0 0 0 0
thank 1 0 1 0 0
will 1 0 2 0 0
Docs
Terms 54605.txt 54642.txt 54660.txt 54662.txt 54664.txt
bit 1 1 1 1 1
document 0 0 0 0 0
folder 0 0 0 0 0
forward 1 0 1 0 0
kean 0 0 0 0 0
pdt 1 1 1 1 1
skeannsf 0 0 0 0 0
steven 0 0 0 0 0
stevenkeanenroncom 0 0 0 0 0
thank 0 2 0 0 0
will 1 1 1 0 0
Docs
Terms 54671.txt 54674.txt 54678.txt 54681.txt 54846.txt
bit 1 1 1 1 1
document 0 0 0 0 0
folder 0 0 0 0 0
forward 0 0 0 0 0
kean 0 0 0 0 0
pdt 1 1 1 1 1
skeannsf 0 0 0 0 0
steven 0 0 0 0 0
stevenkeanenroncom 0 0 0 0 0
thank 0 0 1 0 0
will 0 0 3 0 0
Docs
Terms 55210.txt 65969.txt 9085.txt 9159.txt 9191.txt
bit 1 1 1 1 1
document 0 0 0 0 0
folder 0 0 0 0 0
forward 0 0 0 0 0
kean 0 0 0 0 1
pdt 0 0 1 1 1
skeannsf 0 0 0 0 0
steven 0 0 0 0 0
stevenkeanenroncom 0 0 0 0 0
thank 0 0 0 2 0
will 0 0 3 1 0
Favorite dictionary words
Or we can make make a term-document matrix focusing on words in a dictionary and look at just those columns:
inspect(TermDocumentMatrix(enron_stripped, list(dictionary = c("california", "utah", "texas")))[, 1:5])
<<TermDocumentMatrix (terms: 3, documents: 5)>>
Non-/sparse entries: 2/13
Sparsity : 87%
Maximal term length: 10
Weighting : term frequency (tf)
Docs
Terms 114495.txt 115323.txt 118675.txt 119214.txt 12030.txt
california 27 26 0 0 0
texas 0 0 0 0 0
utah 0 0 0 0 0
Most frequent words
Which terms appear at least 200 times?
findFreqTerms(enron_tdm, 200)
[1] "california" "compani" "contract" "electr" "energi"
[6] "gas" "generat" "market" "new" "plant"
[11] "power" "price" "said" "state" "time"
[16] "util" "will" "year"
Word associations
Which words co-occur frequently with “california”?
findAssocs(enron_tdm, "california", 0.90)
$california
citizen condit
1.00 1.00
consum credit
1.00 1.00
dynegi economist
1.00 1.00
electr energi
1.00 1.00
enough essenti
1.00 1.00
forc higher
1.00 1.00
includ increas
1.00 1.00
investig locat
1.00 1.00
market order
1.00 1.00
power press
1.00 1.00
provid regul
1.00 1.00
reliabl report
1.00 1.00
roll sacramento
1.00 1.00
statewid step
1.00 1.00
structur summer
1.00 1.00
suppli three
1.00 1.00
weather whose
1.00 1.00
abnorm abolish
0.99 0.99
abraham absolut
0.99 0.99
access accomplish
0.99 0.99
accru accumul
0.99 0.99
accus achiev
0.99 0.99
acknowledg acquisit
0.99 0.99
actor acut
0.99 0.99
adam add
0.99 0.99
addit addresse
0.99 0.99
adequaci adjac
0.99 0.99
adjust adopt
0.99 0.99
advantag advers
0.99 0.99
advoc advocaci
0.99 0.99
affect agenc
0.99 0.99
agent aggress
0.99 0.99
aid air
0.99 0.99
airport alamito
0.99 0.99
alarm albani
0.99 0.99
alert aliv
0.99 0.99
allegheni alliant
0.99 0.99
allot allow
0.99 0.99
alon also
0.99 0.99
altern altogeth
0.99 0.99
amass amoco
0.99 0.99
amok ampl
0.99 0.99
analysi angel
0.99 0.99
angelid angus
0.99 0.99
anjali annual
0.99 0.99
anticompetit antitrust
0.99 0.99
antonio anywher
0.99 0.99
appendix appl
0.99 0.99
applianc appointe
0.99 0.99
approv arco
0.99 0.99
argu arguabl
0.99 0.99
aros arrang
0.99 0.99
arrest artifici
0.99 0.99
asid aspect
0.99 0.99
assembl assemblyman
0.99 0.99
assert atkinson
0.99 0.99
attack attorney
0.99 0.99
attribut audra
0.99 0.99
augusta automat
0.99 0.99
avail averag
0.99 0.99
avert await
0.99 0.99
award aybear
0.99 0.99
backup backyard
0.99 0.99
bad bakersfield
0.99 0.99
balanc bangor
0.99 0.99
bankruptci bar
0.99 0.99
barg bargain
0.99 0.99
barn basic
0.99 0.99
basin battin
0.99 0.99
bccd beaten
0.99 0.99
bed begun
0.99 0.99
beloit benchmark
0.99 0.99
benefit berkeley
0.99 0.99
berkshir berkshirehathaway
0.99 0.99
bernadino berthold
0.99 0.99
besid beyer
0.99 0.99
bigger biggest
0.99 0.99
billion biomass
0.99 0.99
bitten black
0.99 0.99
blackout bless
0.99 0.99
blunt blyth
0.99 0.99
bond bonus
0.99 0.99
boost border
0.99 0.99
borderless borenstein
0.99 0.99
born bostonbas
0.99 0.99
bought bounti
0.99 0.99
bowen bowr
0.99 0.99
breach break
0.99 0.99
breakdown brick
0.99 0.99
bridg brink
0.99 0.99
british brka
0.99 0.99
broadbas broadcast
0.99 0.99
brown budhraja
0.99 0.99
buffett built
0.99 0.99
bulk bump
0.99 0.99
bureaucrat burglar
0.99 0.99
burton bustillo
0.99 0.99
bustl caith
0.99 0.99
calenergi calif
0.99 0.99
califbas californialik
0.99 0.99
californiastyl californiatyp
0.99 0.99
calipatria caliso
0.99 0.99
campaign cancel
0.99 0.99
cape carl
0.99 0.99
carolina carri
0.99 0.99
carsonbas cashstrap
0.99 0.99
catch caus
0.99 0.99
caution cec
0.99 0.99
cedar cent
0.99 0.99
centraleast centro
0.99 0.99
chairwoman chamber
0.99 0.99
charg chastis
0.99 0.99
cheapest chevron
0.99 0.99
chose circl
0.99 0.99
cite citi
0.99 0.99
clarifi clash
0.99 0.99
classifi cleanalthough
0.99 0.99
cleaner cleanest
0.99 0.99
clearlak closeddoor
0.99 0.99
coalfir coalgener
0.99 0.99
coast cod
0.99 0.99
cogener cohn
0.99 0.99
cold colleagu
0.99 0.99
column combat
0.99 0.99
combinedcycl come
0.99 0.99
commentari commiss
0.99 0.99
commission committe
0.99 0.99
compani companyalong
0.99 0.99
companyown comparison
0.99 0.99
complac complain
0.99 0.99
compound conceiv
0.99 0.99
concert congener
0.99 0.99
congest connect
0.99 0.99
connel conserv
0.99 0.99
conservationist consolid
0.99 0.99
contend contenti
0.99 0.99
continu contractor
0.99 0.99
contrari contrast
0.99 0.99
controversi convinc
0.99 0.99
cook cooper
0.99 0.99
cop copyright
0.99 0.99
cordova corner
0.99 0.99
cornerston corollari
0.99 0.99
cos cost
0.99 0.99
counterpart counti
0.99 0.99
countri cow
0.99 0.99
cpuc craft
0.99 0.99
crash creditor
0.99 0.99
creditworthi creek
0.99 0.99
crisi cross
0.99 0.99
crossbord crossroad
0.99 0.99
crunch cub
0.99 0.99
cubic curb
0.99 0.99
current custom
0.99 0.99
cut cutback
0.99 0.99
cycl damrel
0.99 0.99
danger darken
0.99 0.99
davi day
0.99 0.99
dboulder dbuxton
0.99 0.99
dcalif deadlin
0.99 0.99
dealbreak dean
0.99 0.99
deast debat
0.99 0.99
debra decri
0.99 0.99
dedic deeper
0.99 0.99
defend del
0.99 0.99
deleg delic
0.99 0.99
delta demand
0.99 0.99
demandsid denounc
0.99 0.99
depart depend
0.99 0.99
deplor deregul
0.99 0.99
design desir
0.99 0.99
despit detmer
0.99 0.99
deulloa devis
0.99 0.99
devolv diametr
0.99 0.99
diann dig
0.99 0.99
dilemma dilig
0.99 0.99
diminish dirtier
0.99 0.99
disappoint disast
0.99 0.99
discharg disclos
0.99 0.99
dismiss dispatch
0.99 0.99
disrupt distant
0.99 0.99
distress district
0.99 0.99
dittmer divers
0.99 0.99
divestitur divis
0.99 0.99
dmarina dollar
0.99 0.99
donal donnelli
0.99 0.99
dorinson doti
0.99 0.99
dow downplay
0.99 0.99
downstat dportland
0.99 0.99
drag dread
0.99 0.99
driven drop
0.99 0.99
dshafter dspringfield
0.99 0.99
dtoddftenergycom duluthbas
0.99 0.99
dwr dyn
0.99 0.99
earli eas
0.99 0.99
easier easili
0.99 0.99
eastern eclips
0.99 0.99
econ ecotek
0.99 0.99
edison effect
0.99 0.99
eix electricgrid
0.99 0.99
electricitygener electricitypurchas
0.99 0.99
element ellingwood
0.99 0.99
embed emit
0.99 0.99
emshwil endors
0.99 0.99
endur energystarv
0.99 0.99
energytrad england
0.99 0.99
enmiti enorm
0.99 0.99
ensur entitl
0.99 0.99
epg equal
0.99 0.99
equat equival
0.99 0.99
ernaut estim
0.99 0.99
even exceed
0.99 0.99
excus execution
0.99 0.99
exempt exodus
0.99 0.99
exoner expans
0.99 0.99
expect expenditur
0.99 0.99
expens expir
0.99 0.99
explan exploit
0.99 0.99
expos extend
0.99 0.99
facil factor
0.99 0.99
fail fairi
0.99 0.99
falloff familiar
0.99 0.99
fan fashion
0.99 0.99
fastest fault
0.99 0.99
feder feet
0.99 0.99
feinstein ferguson
0.99 0.99
ferret fichera
0.99 0.99
fierc fill
0.99 0.99
financ finesalthough
0.99 0.99
fire fiveyear
0.99 0.99
fixedpric flag
0.99 0.99
flat flexibl
0.99 0.99
fli florez
0.99 0.99
fluctuat flux
0.99 0.99
foe fogarti
0.99 0.99
foley foray
0.99 0.99
forecast formula
0.99 0.99
founder fourth
0.99 0.99
fragil franciscobas
0.99 0.99
franck fray
0.99 0.99
fred freed
0.99 0.99
freemarket freez
0.99 0.99
frenzi frequent
0.99 0.99
frustrat fuel
0.99 0.99
full gap
0.99 0.99
gas gascost
0.99 0.99
gasfir gaudett
0.99 0.99
gaug general
0.99 0.99
generat georgia
0.99 0.99
geotherm gerth
0.99 0.99
glenn goal
0.99 0.99
goliath gov
0.99 0.99
gray greater
0.99 0.99
grew grid
0.99 0.99
guarante guidelin
0.99 0.99
gyrat hagerstown
0.99 0.99
hagerstownbas hal
0.99 0.99
half halfown
0.99 0.99
hammer hamper
0.99 0.99
harrigan harsh
0.99 0.99
hathaway hawig
0.99 0.99
headquart healthier
0.99 0.99
heat heavili
0.99 0.99
hebert hefti
0.99 0.99
height herik
0.99 0.99
high highest
0.99 0.99
highlight highpric
0.99 0.99
hike hinik
0.99 0.99
histor hole
0.99 0.99
hopeless horribl
0.99 0.99
hospit household
0.99 0.99
howev httpwwwagostatemaus
0.99 0.99
httpwwwalliantenergycom httpwwwcpuccagov
0.99 0.99
httpwwwisonecom httpwwwmgecom
0.99 0.99
httpwwwnstaronlinecom httpwwwpscstatewius
0.99 0.99
httpwwwwiscuborg httpwwwwisenergycom
0.99 0.99
huard hundr
0.99 0.99
huntington hurri
0.99 0.99
hydro hydroelectr
0.99 0.99
hydroquebec ideolog
0.99 0.99
ignor ill
0.99 0.99
illbas illinoi
0.99 0.99
imag immedi
0.99 0.99
immens immin
0.99 0.99
impair imperi
0.99 0.99
impos imposs
0.99 0.99
impot imprecis
0.99 0.99
inabl inaccur
0.99 0.99
inact inc
0.99 0.99
incent incept
0.99 0.99
incid inclin
0.99 0.99
inconsist indebt
0.99 0.99
independ index
0.99 0.99
indispens ineffect
0.99 0.99
inevit influenti
0.99 0.99
infobox infograph
0.99 0.99
ingram inhibit
0.99 0.99
inject injunct
0.99 0.99
insofar insolv
0.99 0.99
instanc insuffici
0.99 0.99
insul intens
0.99 0.99
interactivecaliforniaimplicationsdoc interactivecaliforniaselecteddoc
0.99 0.99
interconnect interf
0.99 0.99
interfac interjurisdict
0.99 0.99
interpret interrupt
0.99 0.99
interst intervent
0.99 0.99
intrast intrazon
0.99 0.99
introduc investorown
0.99 0.99
involuntari inward
0.99 0.99
iowabas iron
0.99 0.99
irrepar jason
0.99 0.99
jasonleopolddowjonescom jenif
0.99 0.99
jeopardi jersey
0.99 0.99
jerseymaryland jessica
0.99 0.99
jessicabertholddowjonescom jockey
0.99 0.99
joliet jone
0.99 0.99
journalist judg
0.99 0.99
juic justic
0.99 0.99
justif justifi
0.99 0.99
kate keeley
0.99 0.99
keeper kern
0.99 0.99
keynot kid
0.99 0.99
kilowatt kilowatthour
0.99 0.99
king kirk
0.99 0.99
knock lack
0.99 0.99
laid lake
0.99 0.99
lancast landmark
0.99 0.99
landsberg laps
0.99 0.99
larcamp larg
0.99 0.99
larger largest
0.99 0.99
las last
0.99 0.99
lastminut latest
0.99 0.99
lawmak lawrenc
0.99 0.99
lawsuit lawyer
0.99 0.99
legislatur legitim
0.99 0.99
leimgrub leitzing
0.99 0.99
lend length
0.99 0.99
lengthi leopold
0.99 0.99
less lessen
0.99 0.99
leverag levi
0.99 0.99
lie lightbulb
0.99 0.99
lightn likelihood
0.99 0.99
liken line
0.99 0.99
load lobbyist
0.99 0.99
localfeder longer
0.99 0.99
longrang longterm
0.99 0.99
los loss
0.99 0.99
lott low
0.99 0.99
lowest ltd
0.99 0.99
lurch made
0.99 0.99
madison magnitud
0.99 0.99
main major
0.99 0.99
make manhattan
0.99 0.99
manifest manipul
0.99 0.99
map marathon
0.99 0.99
marcel markethedg
0.99 0.99
marketpow marketrestructur
0.99 0.99
maryland mass
0.99 0.99
massachusett massiv
0.99 0.99
mattereveri mayor
0.99 0.99
mcelhenni meanwhil
0.99 0.99
mechan medic
0.99 0.99
mediums megawatt
0.99 0.99
megawatthour meltdown
0.99 0.99
merger merril
0.99 0.99
meter meti
0.99 0.99
mges michaud
0.99 0.99
midamerican midamericancalenergi
0.99 0.99
midatlant midjanuari
0.99 0.99
midst midway
0.99 0.99
midwaysunset midwest
0.99 0.99
miguel millinocket
0.99 0.99
million milwaukeebas
0.99 0.99
minnesota mirror
0.99 0.99
misbehavior mismatch
0.99 0.99
mission mississippian
0.99 0.99
mistak mitchel
0.99 0.99
mitig modest
0.99 0.99
modifi monahan
0.99 0.99
money month
0.99 0.99
morain moratorium
0.99 0.99
morrel mosher
0.99 0.99
motion motiv
0.99 0.99
mount mountain
0.99 0.99
mountainview movement
0.99 0.99
much multipl
0.99 0.99
multipli multiyear
0.99 0.99
municip musel
0.99 0.99
mustrun mwh
0.99 0.99
myer nasti
0.99 0.99
nationwid natur
0.99 0.99
naturalga near
0.99 0.99
nearbankrupt neb
0.99 0.99
need neighbor
0.99 0.99
network nev
0.99 0.99
nevada nevertheless
0.99 0.99
new newli
0.99 0.99
newswir nimock
0.99 0.99
nimockshounorthamericamckinsey nizat
0.99 0.99
nobodi non
0.99 0.99
none nonrmr
0.99 0.99
nontradit normal
0.99 0.99
norman northbrook
0.99 0.99
northeast northeastern
0.99 0.99
northwest notabl
0.99 0.99
notion nrg
0.99 0.99
nstar nuclear
0.99 0.99
nyiso oak
0.99 0.99
obscur observ
0.99 0.99
offici offlin
0.99 0.99
offshor ohio
0.99 0.99
oil okla
0.99 0.99
oklahoma ombudsman
0.99 0.99
omin one
0.99 0.99
onethird oneyear
0.99 0.99
ongo onto
0.99 0.99
open oper
0.99 0.99
oppos opposit
0.99 0.99
ore orga
0.99 0.99
outag outcom
0.99 0.99
outpac output
0.99 0.99
outright outsiz
0.99 0.99
overbuild overcharg
0.99 0.99
overdu overse
0.99 0.99
own oyster
0.99 0.99
pace pacif
0.99 0.99
paid pain
0.99 0.99
pale parallel
0.99 0.99
partial partylin
0.99 0.99
pasadena pass
0.99 0.99
passiv patient
0.99 0.99
pay payer
0.99 0.99
payment pcg
0.99 0.99
peak pennsylvanianew
0.99 0.99
percent period
0.99 0.99
permiss petroleum
0.99 0.99
pge phenomenon
0.99 0.99
picpcx pillar
0.99 0.99
pinnacl pitfal
0.99 0.99
pjm pjms
0.99 0.99
plagu plan
0.99 0.99
plane plant
0.99 0.99
plc pleasant
0.99 0.99
plenti plight
0.99 0.99
plow plug
0.99 0.99
pocket pocta
0.99 0.99
pois pole
0.99 0.99
polic pollut
0.99 0.99
pollutionrel polski
0.99 0.99
pool pore
0.99 0.99
portion pose
0.99 0.99
postpon potenti
0.99 0.99
pound powergener
0.99 0.99
powerget prairi
0.99 0.99
prais pre
0.99 0.99
preced predatori
0.99 0.99
preliminari presum
0.99 0.99
pretextu pri
0.99 0.99
price pricerespons
0.99 0.99
pricesensit principl
0.99 0.99
privatesector pro
0.99 0.99
probe produc
0.99 0.99
profil profit
0.99 0.99
program prolong
0.99 0.99
promin pronounc
0.99 0.99
proof prophet
0.99 0.99
propos pru
0.99 0.99
psc public
0.99 0.99
pull pump
0.99 0.99
punish purchas
0.99 0.99
purpos pursuant
0.99 0.99
push pwr
0.99 0.99
qfs qualifi
0.99 0.99
quarter questar
0.99 0.99
quicken quinn
0.99 0.99
quinta rail
0.99 0.99
rainfal raini
0.99 0.99
rais ram
0.99 0.99
randolph rang
0.99 0.99
rare rate
0.99 0.99
ratepay rattey
0.99 0.99
raw realtim
0.99 0.99
ream reap
0.99 0.99
reason reassur
0.99 0.99
rebat rebuf
0.99 0.99
recent reclaim
0.99 0.99
reconsid recruit
0.99 0.99
red redesign
0.99 0.99
reflect refund
0.99 0.99
regardless region
0.99 0.99
rei reilli
0.99 0.99
reimburs rein
0.99 0.99
reiterman relax
0.99 0.99
relentless relianc
0.99 0.99
reliant reliev
0.99 0.99
remedi renew
0.99 0.99
renobas repaid
0.99 0.99
repair repay
0.99 0.99
repeal repercuss
0.99 0.99
replac replic
0.99 0.99
requir reserv
0.99 0.99
resold resourc
0.99 0.99
restrain result
0.99 0.99
revenu reward
0.99 0.99
rey rhanov
0.99 0.99
rhetor ribbon
0.99 0.99
ridgewood rife
0.99 0.99
rise risen
0.99 0.99
riskavers riski
0.99 0.99
riversid rjacksonvill
0.99 0.99
rla rlancast
0.99 0.99
rmr robinson
0.99 0.99
rock rocket
0.99 0.99
rocki roil
0.99 0.99
rone roosevelt
0.99 0.99
root rose
0.99 0.99
rosemead rough
0.99 0.99
rreno rule
0.99 0.99
runaway runner
0.99 0.99
runup saber
0.99 0.99
sacramentobas said
0.99 0.99
sale salem
0.99 0.99
salt savag
0.99 0.99
save saxl
0.99 0.99
say scenario
0.99 0.99
schoenherr school
0.99 0.99
scof scrambl
0.99 0.99
scrutini secondlargest
0.99 0.99
secreci section
0.99 0.99
seek segundo
0.99 0.99
sell sempra
0.99 0.99
separ set
0.99 0.99
seven sever
0.99 0.99
severin sharehold
0.99 0.99
sheffrin shelv
0.99 0.99
sheppard shift
0.99 0.99
shipment shipper
0.99 0.99
shock shortag
0.99 0.99
shorten shortfal
0.99 0.99
shot show
0.99 0.99
shrink shrunk
0.99 0.99
shut shutdown
0.99 0.99
sierra sign
0.99 0.99
signal similar
0.99 0.99
simpl simpli
0.99 0.99
simultan sinc
0.99 0.99
site sixmonth
0.99 0.99
size skeptic
0.99 0.99
skygen slack
0.99 0.99
slate slight
0.99 0.99
slip sloppi
0.99 0.99
slower small
0.99 0.99
smallbusi smallest
0.99 0.99
smarter smog
0.99 0.99
smooth soar
0.99 0.99
socal softped
0.99 0.99
sokol solar
0.99 0.99
sold solv
0.99 0.99
solvenc sophist
0.99 0.99
sorensen south
0.99 0.99
southern southland
0.99 0.99
spar specialist
0.99 0.99
specifi specter
0.99 0.99
specul spencer
0.99 0.99
spent spike
0.99 0.99
spiral spokesman
0.99 0.99
spokesmen spokeswoman
0.99 0.99
spotmarket spring
0.99 0.99
spur sre
0.99 0.99
stagger stall
0.99 0.99
standard state
0.99 0.99
stateappoint station
0.99 0.99
statist stave
0.99 0.99
steadili steam
0.99 0.99
steep steepli
0.99 0.99
stem stengel
0.99 0.99
stern stick
0.99 0.99
stifl stop
0.99 0.99
storag strain
0.99 0.99
stride strip
0.99 0.99
struggl stunt
0.99 0.99
subsidi subsidiari
0.99 0.99
substanti suffer
0.99 0.99
suffici summar
0.99 0.99
summerlik summit
0.99 0.99
sunset superior
0.99 0.99
supplier surgic
0.99 0.99
surplus surround
0.99 0.99
suspend suspicion
0.99 0.99
sustain suzann
0.99 0.99
suzannenimocksmckinseycom swelter
0.99 0.99
switch sympath
0.99 0.99
system tab
0.99 0.99
tailor tamaki
0.99 0.99
tap taper
0.99 0.99
taxpay tem
0.99 0.99
temperatur tempest
0.99 0.99
temporari temporarili
0.99 0.99
tend tenuous
0.99 0.99
tep testifi
0.99 0.99
testimoni therm
0.99 0.99
thermal thermo
0.99 0.99
third thorni
0.99 0.99
threshold thursday
0.99 0.99
tie time
0.99 0.99
timeconsum tini
0.99 0.99
togeth token
0.99 0.99
toll tool
0.99 0.99
townsend trace
0.99 0.99
trade transenergi
0.99 0.99
translat transmit
0.99 0.99
transplant tread
0.99 0.99
treasur trend
0.99 0.99
trent trim
0.99 0.99
tuck tulsa
0.99 0.99
tulsabas turmoil
0.99 0.99
twice two
0.99 0.99
twothird typic
0.99 0.99
ultim unavail
0.99 0.99
uncertain uncollect
0.99 0.99
undercollect underestim
0.99 0.99
underscor underwat
0.99 0.99
undesir undo
0.99 0.99
undu unfair
0.99 0.99
unfett unfil
0.99 0.99
unforeseen uniform
0.99 0.99
unintend union
0.99 0.99
unlaw unlik
0.99 0.99
unpaid unpreced
0.99 0.99
unregul unseason
0.99 0.99
unwarr upset
0.99 0.99
upstat upward
0.99 0.99
utah util
0.99 0.99
utilitiespacif utilitiessan
0.99 0.99
utilitiesstrenu utilityderegul
0.99 0.99
valley van
0.99 0.99
vanech variabl
0.99 0.99
varieti various
0.99 0.99
vast vegasbas
0.99 0.99
vehement venkateshwara
0.99 0.99
venki verbatim
0.99 0.99
verg veteran
0.99 0.99
victori vigor
0.99 0.99
vikram villaraigosa
0.99 0.99
violat virginia
0.99 0.99
virtu vocal
0.99 0.99
vogel volatil
0.99 0.99
voltag volum
0.99 0.99
vulner wagon
0.99 0.99
wake walli
0.99 0.99
warn warrant
0.99 0.99
warren weak
0.99 0.99
wednesday weeklysometim
0.99 0.99
wellequip wellhead
0.99 0.99
whenev wherea
0.99 0.99
whether whitewat
0.99 0.99
wholeheart widen
0.99 0.99
widerang wild
0.99 0.99
wilkerson will
0.99 0.99
william windfal
0.99 0.99
winter winwin
0.99 0.99
wisconsin wisvest
0.99 0.99
wit within
0.99 0.99
wmb wood
0.99 0.99
worri wors
0.99 0.99
worsen writer
0.99 0.99
wrongdo wwwlatimescompow
0.99 0.99
wyom year
0.99 0.99
yearend yorker
0.99 0.99
zero ziegler
0.99 0.99
abus accord
0.98 0.98
act agreement
0.98 0.98
although among
0.98 0.98
amount april
0.98 0.98
avoid basi
0.98 0.98
beach began
0.98 0.98
bid big
0.98 0.98
build buy
0.98 0.98
capac claim
0.98 0.98
competit construct
0.98 0.98
contract contribut
0.98 0.98
court critic
0.98 0.98
debt decemb
0.98 0.98
decis declin
0.98 0.98
differ done
0.98 0.98
due edit
0.98 0.98
emerg environ
0.98 0.98
examin exchang
0.98 0.98
execut far
0.98 0.98
februari file
0.98 0.98
first five
0.98 0.98
focus foster
0.98 0.98
fulli give
0.98 0.98
hand hot
0.98 0.98
hour huge
0.98 0.98
impact import
0.98 0.98
industri issu
0.98 0.98
januari keep
0.98 0.98
known law
0.98 0.98
lose lot
0.98 0.98
lower mani
0.98 0.98
march margin
0.98 0.98
marketplac may
0.98 0.98
measur memorandum
0.98 0.98
miller monitor
0.98 0.98
must negoti
0.98 0.98
novemb often
0.98 0.98
owe owner
0.98 0.98
part parti
0.98 0.98
paso possibl
0.98 0.98
problem process
0.98 0.98
promis prompt
0.98 0.98
puc refus
0.98 0.98
regulatori relat
0.98 0.98
repres request
0.98 0.98
right run
0.98 0.98
serious servic
0.98 0.98
signific situat
0.98 0.98
sourc spot
0.98 0.98
stabil staff
0.98 0.98
still suggest
0.98 0.98
supervisor though
0.98 0.98
took total
0.98 0.98
toward transact
0.98 0.98
troubl unit
0.98 0.98
use want
0.98 0.98
water western
0.98 0.98
without yet
0.98 0.98
york acceler
0.98 0.97
actual administr
0.97 0.97
agre ahead
0.97 0.97
almost alreadi
0.97 0.97
appeal applic
0.97 0.97
associ audit
0.97 0.97
base calpin
0.97 0.97
came can
0.97 0.97
case center
0.97 0.97
certain commerc
0.97 0.97
conclud constraint
0.97 0.97
consult correct
0.97 0.97
cover deal
0.97 0.97
decad declar
0.97 0.97
determin economi
0.97 0.97
effici effort
0.97 0.97
employ enter
0.97 0.97
everi excess
0.97 0.97
expand express
0.97 0.97
face fall
0.97 0.97
figur financi
0.97 0.97
flaw fpa
0.97 0.97
home implic
0.97 0.97
insid instead
0.97 0.97
interest lead
0.97 0.97
like might
0.97 0.97
monday monopoli
0.97 0.97
net news
0.97 0.97
next nonpubl
0.97 0.97
oregon overal
0.97 0.97
per pipelin
0.97 0.97
posit practic
0.97 0.97
privat properti
0.97 0.97
prove qualiti
0.97 0.97
refer remain
0.97 0.97
resid rest
0.97 0.97
share sought
0.97 0.97
submit threaten
0.97 0.97
thus told
0.97 0.97
tri twoyear
0.97 0.97
user way
0.97 0.97
won across
0.97 0.96
address ancillari
0.96 0.96
away begin
0.96 0.96
behind breath
0.96 0.96
bring buyer
0.96 0.96
call ceil
0.96 0.96
chang clear
0.96 0.96
collect complet
0.96 0.96
complex condition
0.96 0.96
consequ coordin
0.96 0.96
corp direct
0.96 0.96
director duti
0.96 0.96
earlier emiss
0.96 0.96
form futur
0.96 0.96
get given
0.96 0.96
hard hit
0.96 0.96
identifi illeg
0.96 0.96
infrastructur inquiri
0.96 0.96
just late
0.96 0.96
later long
0.96 0.96
matter mean
0.96 0.96
necessarili northern
0.96 0.96
procedur procur
0.96 0.96
put receiv
0.96 0.96
respect respons
0.96 0.96
rush see
0.96 0.96
solut someth
0.96 0.96
spread stage
0.96 0.96
stand tariff
0.96 0.96
text therefor
0.96 0.96
tight top
0.96 0.96
trader tuesday
0.96 0.96
turn unless
0.96 0.96
withhold worth
0.96 0.96
ago anoth
0.95 0.95
articl author
0.95 0.95
californian consider
0.95 0.95
daili data
0.95 0.95
describ develop
0.95 0.95
end experienc
0.95 0.95
happen inform
0.95 0.95
intern interview
0.95 0.95
joseph kept
0.95 0.95
marketbas offer
0.95 0.95
panel proceed
0.95 0.95
product project
0.95 0.95
republican sometim
0.95 0.95
spend statement
0.95 0.95
texa track
0.95 0.95
transport wall
0.95 0.95
week west
0.95 0.95
afford alloc
0.94 0.94
along appear
0.94 0.94
approach assess
0.94 0.94
burden calcul
0.94 0.94
cash civil
0.94 0.94
club coal
0.94 0.94
compar compens
0.94 0.94
connecticut definit
0.94 0.94
easi entir
0.94 0.94
entiti eventu
0.94 0.94
find former
0.94 0.94
free georg
0.94 0.94
insist investor
0.94 0.94
jay labor
0.94 0.94
launch llc
0.94 0.94
manner mile
0.94 0.94
night occur
0.94 0.94
onlin parent
0.94 0.94
pend peopl
0.94 0.94
persuad point
0.94 0.94
polit politician
0.94 0.94
portfolio previous
0.94 0.94
print progress
0.94 0.94
question retroact
0.94 0.94
settl sharp
0.94 0.94
short side
0.94 0.94
split strict
0.94 0.94
suppos suspect
0.94 0.94
swept take
0.94 0.94
term transform
0.94 0.94
wind work
0.94 0.94
abil asset
0.93 0.93
brought compet
0.93 0.93
default deliv
0.93 0.93
elsewher especi
0.93 0.93
exact fix
0.93 0.93
foundat individu
0.93 0.93
influenc island
0.93 0.93
jan legisl
0.93 0.93
light limit
0.93 0.93
maintain nonprofit
0.93 0.93
note partner
0.93 0.93
primari reform
0.93 0.93
role secur
0.93 0.93
sen taken
0.93 0.93
today ann
0.93 0.92
becam bill
0.92 0.92
bush cap
0.92 0.92
comment detail
0.92 0.92
difficult fair
0.92 0.92
fee institut
0.92 0.92
involv iso
0.92 0.92
local met
0.92 0.92
move number
0.92 0.92
past place
0.92 0.92
presid publish
0.92 0.92
recommend record
0.92 0.92
specif tax
0.92 0.92
well allen
0.92 0.91
believ commod
0.91 0.91
communiti dan
0.91 0.91
disclosur encourag
0.91 0.91
error evid
0.91 0.91
ferc global
0.91 0.91
gone initi
0.91 0.91
legal lynch
0.91 0.91
materi meantim
0.91 0.91
review sat
0.91 0.91
soon sound
0.91 0.91
start talk
0.91 0.91
truth washington
0.91 0.91
announc attempt
0.90 0.90
commit concern
0.90 0.90
deni enforc
0.90 0.90
expedit extens
0.90 0.90
hold morgan
0.90 0.90
other prevent
0.90 0.90
provis reaction
0.90 0.90
reduc schedul
0.90 0.90
stay street
0.90 0.90
unabl
0.90
What else might you do?
- Use the
tidytextpackage to work in a “tidy” way - Make more visualizations of word frequencies or relationships in
ggplot2 - Use hierarchical clustering to group together terms
- Fit topic models to find overarching topics
- Use
NLPpackage to find bigrams (two-word phrases) - Use
qdappackage to classify document sentiment (positive, negative) and look for relationships
(Please social science responsibly!)
Wrapping up the course
What you've learned
A lot!
- How to get data into R from a variety of formats
- How to do “data janitor” work to manipulate and clean data
- How to make pretty visualizations
- How to automate with loops and functions
- How to fit linear regression models
- How to combine text, calculations, plots, and tables into dynamic R Markdown reports
What comes next?
- Doing statistical inference
- Functions for hypothesis testing, hierarchical/mixed effect models, machine learning, survey design, etc. straightforward to use…once data are clean
- Access output by working with list structures (like with linear regression HW)
- Practice, practice, practice!
- Replicate analyses you've done in Excel, SPSS, Stata
- Think about data using
dplyrverbs, tidy data principles - R Markdown for documenting cleaning and analysis start-to-finish
- More advanced
- Using version control (git) in RStudio
- Interactive Shiny web apps