In trinker/qdap: Bridging the Gap Between Qualitative Data and Quantitative Analysis
&twocol
# set global chunk options
library(reports); library(slidify); library(knitcitations); library(knitr); library(qdap)
opts_chunk$set(cache=FALSE, tidy = FALSE)
#Reading in your own bib file:
bib <- read.bibtex(dir()[tools::file_ext(dir()) == "bib"][1])
#cite in text using `r citet(bib[1])`
internal <- TRUE
About Me
*** =right
r VS(2)
- Former First Grade Teacher
- Literacy PhD Student
- Quantitatively Bent
- Accidental Programmer
*** =left
r IM2("norah.png", width=350)
*** =pnotes
Research Interests
- Engagement, Motivation and Feedback
- Multimodal Analysis
- Discourse Analysis
- Data Visualization
- Improving Research Methods
--- &twocol
Why R?
*** =left
r VS(2)
- Cutting Edge
- Powerful
- Visualization
r VS(4) r HS(4) Everything can be quantified...
*** =right
r VS(2)
r IM2("r.jpg", width=350)
--- {class: class, tpl: tabs}
Why qdap?
*** {class: active, id: qdap}
r IM2("whyqdap.png")
*** {id: Birth}
Frustration
r IM2("fustration.jpg", width = 400)
*** {id: Affordances}
Affordances 
- Fexible
-
Bridge to r HR("http://cran.r-project.org/", "5087") + r HR("http://www.bioconductor.org/", "749") + r HR("https://github.com", "?") = +5836
-
Dynamic (Field Driven)
- Data Focused
Recomendations
r IM2("logo_rstudio.jpg", width = 200, center = FALSE, link = "http://www.rstudio.com/") r IM2("github-social.png", width = 200, center = FALSE, link ="https://github.com/") r IM2("knitr.png", width = 200, center = FALSE, link ="http://yihui.name/knitr/") + r IM2("latex.png", width = 160, center = FALSE, link = "http://en.wikipedia.org/wiki/LaTeX") r HS(5) r HR("C:/Users/trinker/Desktop/proximity/REPORT/proximity_measure.Rnw", ".Rmd") r HS(5)``r HR("C:/Users/trinker/Desktop/proximity/REPORT/proximity_measure.pdf", ".pdf")
r IM2("ggplot.png", width = 200, center = FALSE, link = "http://docs.ggplot2.org/current/")
press p
*** =pnotes
r IM2("ggplot2demo.png")
Let's Dig In
r IM2("dig.jpg")
*** =pnotes
Agenda
- Installing qdap
- Projects
- Read In Data
- Word Counts and Descriptive Statistics
- Word Measures and Scoring
- Qualitative Coding System
- Visualizing Discourse Data
- Discussion
Installing qdap
# install.packages("devtools", eval = FALSE)
library(devtools)
install_github(c("slidify", "slidifyLibraries"), "ramnathv", ref = "dev")
install_github("knitcitations", "cboettig")
install_github(c("reports", "qdapDictionaries", "qdap"), "trinker")
install_github("ggthemes", "jrnold")
install.packages("scales")
invisible(lapply(c("qdap", "ggplot2", "ggthemes", "scales", "grid"),
require, character.only = TRUE))
r VS(2)
https://github.com/trinker/qdap
--- .YT yt:chQlpEj8g2Q &youtube
Projects
*** =pnotes
r IF("vid1.html")
General qdap Function Format
Function(Text_Variable, list(Grouping_Variables))
r VS(2)
with(Data_Set, Function(Text_Variable, list(Grouping_Variables)))
Read In Data
- Word (docx)
- Text (txt)
- Excel (csv/xlsx)
Read In Data
r IM2("transcript.png", width= 850)
Read In Data
doc1 <- system.file("extdata/transcripts/trans1.docx", package = "qdap")
dat1 <- read.transcript(doc1)
truncdf(dat1, 50)
r VS(2)
Plenty of parsing tools to clean up!!!
Our Data Set
DATA
suppressMessages(library(xtable))
print(xtable(DATA), type="html")
Word Counts and Descriptive Statistics
- Word Frequency Matrix
- Word Stats
- Term Counts
- Question Types
- Parts of Speech
- Syllablication
Word Frequency Matrix
with(DATA, wfm(state, person))[1:14, ]
Word Frequency Matrix
plot(with(DATA, wfm(state, person)), values = TRUE, plot = FALSE) +
coord_flip()
Word Frequency Matrix (Correlations)
dat2 <- wfm(DATA$state, seq_len(nrow(DATA)))
qheat(cor(t(dat2)), low = "yellow", high = "red",
grid = "grey90", diag.na = TRUE, by.column = NULL)
Word Stats (1 of 3)
(desc_wrds <- with(mraja1spl, word_stats(dialogue, person, tot = tot)))
desc_wrds2 <- with(mraja1spl, word_stats(desc_wrds, person, tot = tot, digits = 1))
desc_wrds2$gts[, c(1, 2:9)]
Word Stats (2 of 3)
desc_wrds2$gts[, c(1, 10:19)]
Word Stats (3 of 3)
desc_wrds2$gts[, c(1, 20:26)]
Word Stats Plot
plot(desc_wrds, label = TRUE, high="red")
Term Counts
ml2 <- list(
theme_1 = c(" the ", " a ", " an "),
theme_2 = c(" I'" ),
"good",
the_words = c("the", " the ", " the", "the ")
)
out <- with(raj.act.1, termco(dialogue, person, ml2))
*** =pnotes
r VS(2) *Press p
ml2 <- list(
theme_1 = c(" the ", " a ", " an "),
theme_2 = c(" I'" ),
"good",
the_words = c("the", " the ", " the", "the ")
)
Term Counts
out
*** =pnotes
ml2 <- list(
theme_1 = c(" the ", " a ", " an "),
theme_2 = c(" I'" ),
"good",
the_words = c("the", " the ", " the", "the ")
)
Term Counts Plot
plot(out, high = "red", low = "yellow", label = TRUE)
Question Types
(x <- with(mraja1spl, question_type(dialogue, person)))
Question Types Plot
plot(x)
Parts of Speech
(posbydat <- with(DATA, pos_by(state, list(adult, sex))))
load("pos.RData")
posbydat
r VS(3) r HR("http://faculty.washington.edu/dillon/GramResources/penntable.html", "Penn Treebank Project (1991)")
*** =pnotes
posbydat[["POStagged"]]
Parts of Speech Plot
plot(posbydat, label = TRUE)
Word Measures and Scoring
- Readability
- Formality
- Polarity
Readability
- Automated Readability Index
- Coleman Liau
- SMOG
- Flesch Kincaid
- Fry
- Linsear Write
Readability
with(rajSPLIT, coleman_liau(dialogue, list(fam.aff)))
Formality
Heylighen & Dewaele(1999a, 1999b, 2002)
$$ F = 50(\frac{n_{f}-n_{c}}{N} + 1) $$
Where:
$$ f = \left {noun, \;adjective, \;preposition, \;article\right } $$
$$ c = \left {pronoun, \;verb, \;adverb, \;interjection\right } $$
$$ N = \sum{(f \;+ \;c \;+ \;conjunctions)} $$
Formality
(form <- with(raj, formality(dialogue, act)))
(form <- with(raj, formality(rajPOS, act)))
Formality Plot
plot(form, bar.colors=c("Set2", "RdBu"))
Polarity
(poldat <- with(mraja1spl, polarity(dialogue, list(sex, fam.aff, died))))
plot(poldat)
Qualitative Coding System
Apply codes to:
- The dialogue
- The time spans
codes <- qcv(AA, BB, CC)
X <- cm_df.transcript(DATA$state, DATA$person, file="DATA.txt")
sam:
1 2 3 4 5 6
Computer is fun. Not too fun.
greg:
7 8 9 10 11
No it's not, it's dumb.
teacher:
12 13 14 15
What should we do?
sam:
16 17 18 19
You liar, it stinks!
r IM2("transcript2.png", width=800)
r VS(1)
Coding time spans works similarly...
press p
*** =pnotes
x <- list(
transcript_time_span = qcv(00:00 - 1:12:00),
A = qcv(terms = "2.40:3.00, 5.01, 6.02:7.00, 9.00,
12.00:14.00, 00.51.00:00.59.00"),
B = qcv(terms = "2.40, 3.01:3.02, 5.01, 6.02:7.00,
9.00, 30.00:39.00, 1.12.00:1.19.01"),
C = qcv(terms = "2.40:3.00, 5.01, 6.02:7.00, 9.00, 17.01")
)
After Reading Qualitative Codes...
(y <- cm_2long(x))
Gantt Plot of Codes
plot(y)
Summary of Codes
summary(y)
plot(summary(y))
Visualizing Discourse Data
- Lexical Dispersion Plot
- Word Cloud
- Turn of Talk Plot
- Venn Diagram
- Word Network Plot
Lexical Dispersion Plot
with(rajSPLIT , dispersion_plot(dialogue, c("love", "night"),
grouping.var = list(fam.aff, sex), rm.vars = act))
Lexical Dispersion Plot
wrds <- word_list(pres_debates2012$dialogue, stopwords = Top200Words)
wrds2 <- spaste(wrds[["rfswl"]][["all"]][, "WORD"])
wrds2 <- c(" governor~~romney ", wrds2[-c(3, 12)])
with(pres_debates2012 , dispersion_plot(dialogue, wrds2, rm.vars = time, color="black", bg.color="white"))
Word Cloud (Colored Terms)
terms <- list(
I = c("i", "i'm"),
mal = qcv(stinks, dumb, distrust),
articles = qcv(the, a, an),
pronoun = qcv(we, you)
)
with(DATA, trans_cloud(state, target.words=terms,
cloud.colors=qcv(red, green, blue, black, gray65),
expand.target=FALSE, proportional=TRUE, legend=c(names(terms),
"other")))
terms <- list(
I = c("i", "i'm"),
mal = qcv(stinks, dumb, distrust),
articles = qcv(the, a, an),
pronoun = qcv(we, you)
)
with(DATA, trans_cloud(state, target.words=terms,
cloud.colors=qcv(red, green, blue, black, gray65),
expand.target=FALSE, proportional=TRUE, legend=c(names(terms),
"other")))
Word Cloud Plot (Colored Terms)
with(DATA, trans_cloud(state, target.words=terms,
cloud.colors=qcv(red, green, blue, black, gray65),
expand.target=FALSE, proportional=TRUE, legend=c(names(terms),
"other"), max.word.size = 8, min.word.size = .5))
Gradient Cloud
DATA2 <- DATA
DATA2$state <- space_fill(DATA$state, c("is fun", "too fun", "you liar"))
gradient_cloud(DATA$state, DATA$sex, title="Houghton Colors",
max.word.size = 8, min.word.size = .01, X ="purple" , Y = "yellow")
Gradient Cloud
gradient_cloud(DATA2$state, DATA2$sex, title="Gender Word Use",
max.word.size = 8, min.word.size = .01, X ="red" , Y = "blue")
Turn of Talk Plot
r VS(1)
tot_plot(mraja1, "dialogue", grouping.var = c("sex", "fam.aff"), tot=FALSE, plot=FALSE)+
scale_fill_brewer(palette="Set1") +
geom_hline(aes(yintercept=mean(word.count))) +
geom_hline(aes(yintercept=mean(word.count) + (2 *sd(word.count)))) +
geom_hline(aes(yintercept=mean(word.count) + (3 *sd(word.count)))) +
geom_text(parse=TRUE, hjust=0, vjust=0, size = 3, aes(x = 2,
y = mean(word.count) + 2, label = "bar(x)")) +
geom_text(hjust=0, vjust=0, size = 3, aes(x = 1,
y = mean(word.count) + (2 *sd(word.count)) + 2, label = "+2 sd")) +
geom_text(hjust=0, vjust=0, size = 3, aes(x = 1,
y = mean(word.count) + (3 *sd(word.count)) + 2, label = "+3 sd")) +
ggtitle("Romeo & Juliet:\nAct 1 Turns of Talk")
Venn Diagram
with(DATA , trans_venn(state, person, legend.location = "topright"))
Word Network Plot
word_network_plot(text.var=DATA$state, DATA$person, stopwords=NULL)
Discussion
- How might qdap + R fit into your workflow?
- What do you want to know more about?
- Are there any points that need to be clarified?
- …
trinker/qdap documentation built on Sept. 30, 2020, 6:28 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
&twocol
# set global chunk options library(reports); library(slidify); library(knitcitations); library(knitr); library(qdap) opts_chunk$set(cache=FALSE, tidy = FALSE) #Reading in your own bib file: bib <- read.bibtex(dir()[tools::file_ext(dir()) == "bib"][1]) #cite in text using `r citet(bib[1])` internal <- TRUE
About Me
*** =right
r VS(2)
- Former First Grade Teacher
- Literacy PhD Student
- Quantitatively Bent
- Accidental Programmer
*** =left
r IM2("norah.png", width=350)
*** =pnotes
Research Interests
- Engagement, Motivation and Feedback
- Multimodal Analysis
- Discourse Analysis
- Data Visualization
- Improving Research Methods
--- &twocol
Why R?
*** =left
r VS(2)
- Cutting Edge
- Powerful
- Visualization
r VS(4) r HS(4) Everything can be quantified...
*** =right
r VS(2)
r IM2("r.jpg", width=350)
--- {class: class, tpl: tabs}
Why qdap?
*** {class: active, id: qdap}
r IM2("whyqdap.png")
*** {id: Birth}
Frustration
r IM2("fustration.jpg", width = 400)
*** {id: Affordances}
Affordances
- Fexible
-
Bridge to
r HR("http://cran.r-project.org/", "5087")+r HR("http://www.bioconductor.org/", "749")+r HR("https://github.com", "?")= +5836 -
Dynamic (Field Driven)
- Data Focused
Recomendations
r IM2("logo_rstudio.jpg", width = 200, center = FALSE, link = "http://www.rstudio.com/")r IM2("github-social.png", width = 200, center = FALSE, link ="https://github.com/")r IM2("knitr.png", width = 200, center = FALSE, link ="http://yihui.name/knitr/")+r IM2("latex.png", width = 160, center = FALSE, link = "http://en.wikipedia.org/wiki/LaTeX")r HS(5)r HR("C:/Users/trinker/Desktop/proximity/REPORT/proximity_measure.Rnw", ".Rmd")r HS(5)``r HR("C:/Users/trinker/Desktop/proximity/REPORT/proximity_measure.pdf", ".pdf")r IM2("ggplot.png", width = 200, center = FALSE, link = "http://docs.ggplot2.org/current/")
press p
*** =pnotes
r IM2("ggplot2demo.png")
Let's Dig In
r IM2("dig.jpg")
*** =pnotes
Agenda
- Installing qdap
- Projects
- Read In Data
- Word Counts and Descriptive Statistics
- Word Measures and Scoring
- Qualitative Coding System
- Visualizing Discourse Data
- Discussion
Installing qdap
# install.packages("devtools", eval = FALSE)
library(devtools)
install_github(c("slidify", "slidifyLibraries"), "ramnathv", ref = "dev")
install_github("knitcitations", "cboettig")
install_github(c("reports", "qdapDictionaries", "qdap"), "trinker")
install_github("ggthemes", "jrnold")
install.packages("scales")
invisible(lapply(c("qdap", "ggplot2", "ggthemes", "scales", "grid"),
require, character.only = TRUE))
r VS(2)
https://github.com/trinker/qdap
--- .YT yt:chQlpEj8g2Q &youtube
Projects
*** =pnotes
r IF("vid1.html")
General qdap Function Format
Function(Text_Variable, list(Grouping_Variables))
r VS(2)
with(Data_Set, Function(Text_Variable, list(Grouping_Variables)))
Read In Data
- Word (docx)
- Text (txt)
- Excel (csv/xlsx)
Read In Data
r IM2("transcript.png", width= 850)
Read In Data
doc1 <- system.file("extdata/transcripts/trans1.docx", package = "qdap") dat1 <- read.transcript(doc1) truncdf(dat1, 50)
r VS(2)
Plenty of parsing tools to clean up!!!
Our Data Set
DATA
suppressMessages(library(xtable)) print(xtable(DATA), type="html")
Word Counts and Descriptive Statistics
- Word Frequency Matrix
- Word Stats
- Term Counts
- Question Types
- Parts of Speech
- Syllablication
Word Frequency Matrix
with(DATA, wfm(state, person))[1:14, ]
Word Frequency Matrix
plot(with(DATA, wfm(state, person)), values = TRUE, plot = FALSE) + coord_flip()
Word Frequency Matrix (Correlations)
dat2 <- wfm(DATA$state, seq_len(nrow(DATA))) qheat(cor(t(dat2)), low = "yellow", high = "red", grid = "grey90", diag.na = TRUE, by.column = NULL)
Word Stats (1 of 3)
(desc_wrds <- with(mraja1spl, word_stats(dialogue, person, tot = tot)))
desc_wrds2 <- with(mraja1spl, word_stats(desc_wrds, person, tot = tot, digits = 1))
desc_wrds2$gts[, c(1, 2:9)]
Word Stats (2 of 3)
desc_wrds2$gts[, c(1, 10:19)]
Word Stats (3 of 3)
desc_wrds2$gts[, c(1, 20:26)]
Word Stats Plot
plot(desc_wrds, label = TRUE, high="red")
Term Counts
ml2 <- list( theme_1 = c(" the ", " a ", " an "), theme_2 = c(" I'" ), "good", the_words = c("the", " the ", " the", "the ") )
out <- with(raj.act.1, termco(dialogue, person, ml2))
*** =pnotes
r VS(2) *Press p
ml2 <- list( theme_1 = c(" the ", " a ", " an "), theme_2 = c(" I'" ), "good", the_words = c("the", " the ", " the", "the ") )
Term Counts
out
*** =pnotes
ml2 <- list( theme_1 = c(" the ", " a ", " an "), theme_2 = c(" I'" ), "good", the_words = c("the", " the ", " the", "the ") )
Term Counts Plot
plot(out, high = "red", low = "yellow", label = TRUE)
Question Types
(x <- with(mraja1spl, question_type(dialogue, person)))
Question Types Plot
plot(x)
Parts of Speech
(posbydat <- with(DATA, pos_by(state, list(adult, sex))))
load("pos.RData") posbydat
r VS(3) r HR("http://faculty.washington.edu/dillon/GramResources/penntable.html", "Penn Treebank Project (1991)")
*** =pnotes
posbydat[["POStagged"]]
Parts of Speech Plot
plot(posbydat, label = TRUE)
Word Measures and Scoring
- Readability
- Formality
- Polarity
Readability
- Automated Readability Index
- Coleman Liau
- SMOG
- Flesch Kincaid
- Fry
- Linsear Write
Readability
with(rajSPLIT, coleman_liau(dialogue, list(fam.aff)))
Formality
Heylighen & Dewaele(1999a, 1999b, 2002)
$$ F = 50(\frac{n_{f}-n_{c}}{N} + 1) $$
Where:
$$ f = \left {noun, \;adjective, \;preposition, \;article\right } $$
$$ c = \left {pronoun, \;verb, \;adverb, \;interjection\right } $$
$$ N = \sum{(f \;+ \;c \;+ \;conjunctions)} $$
Formality
(form <- with(raj, formality(dialogue, act)))
(form <- with(raj, formality(rajPOS, act)))
Formality Plot
plot(form, bar.colors=c("Set2", "RdBu"))
Polarity
(poldat <- with(mraja1spl, polarity(dialogue, list(sex, fam.aff, died))))
plot(poldat)
Qualitative Coding System
Apply codes to:
- The dialogue
- The time spans
codes <- qcv(AA, BB, CC)
X <- cm_df.transcript(DATA$state, DATA$person, file="DATA.txt")
sam:
1 2 3 4 5 6
Computer is fun. Not too fun.
greg:
7 8 9 10 11
No it's not, it's dumb.
teacher:
12 13 14 15
What should we do?
sam:
16 17 18 19
You liar, it stinks!
r IM2("transcript2.png", width=800)
r VS(1)
Coding time spans works similarly...
press p
*** =pnotes
x <- list( transcript_time_span = qcv(00:00 - 1:12:00), A = qcv(terms = "2.40:3.00, 5.01, 6.02:7.00, 9.00, 12.00:14.00, 00.51.00:00.59.00"), B = qcv(terms = "2.40, 3.01:3.02, 5.01, 6.02:7.00, 9.00, 30.00:39.00, 1.12.00:1.19.01"), C = qcv(terms = "2.40:3.00, 5.01, 6.02:7.00, 9.00, 17.01") )
After Reading Qualitative Codes...
(y <- cm_2long(x))
Gantt Plot of Codes
plot(y)
Summary of Codes
summary(y) plot(summary(y))
Visualizing Discourse Data
- Lexical Dispersion Plot
- Word Cloud
- Turn of Talk Plot
- Venn Diagram
- Word Network Plot
Lexical Dispersion Plot
with(rajSPLIT , dispersion_plot(dialogue, c("love", "night"), grouping.var = list(fam.aff, sex), rm.vars = act))
Lexical Dispersion Plot
wrds <- word_list(pres_debates2012$dialogue, stopwords = Top200Words) wrds2 <- spaste(wrds[["rfswl"]][["all"]][, "WORD"]) wrds2 <- c(" governor~~romney ", wrds2[-c(3, 12)]) with(pres_debates2012 , dispersion_plot(dialogue, wrds2, rm.vars = time, color="black", bg.color="white"))
Word Cloud (Colored Terms)
terms <- list(
I = c("i", "i'm"),
mal = qcv(stinks, dumb, distrust),
articles = qcv(the, a, an),
pronoun = qcv(we, you)
)
with(DATA, trans_cloud(state, target.words=terms,
cloud.colors=qcv(red, green, blue, black, gray65),
expand.target=FALSE, proportional=TRUE, legend=c(names(terms),
"other")))
terms <- list( I = c("i", "i'm"), mal = qcv(stinks, dumb, distrust), articles = qcv(the, a, an), pronoun = qcv(we, you) ) with(DATA, trans_cloud(state, target.words=terms, cloud.colors=qcv(red, green, blue, black, gray65), expand.target=FALSE, proportional=TRUE, legend=c(names(terms), "other")))
Word Cloud Plot (Colored Terms)
with(DATA, trans_cloud(state, target.words=terms, cloud.colors=qcv(red, green, blue, black, gray65), expand.target=FALSE, proportional=TRUE, legend=c(names(terms), "other"), max.word.size = 8, min.word.size = .5))
Gradient Cloud
DATA2 <- DATA DATA2$state <- space_fill(DATA$state, c("is fun", "too fun", "you liar"))
gradient_cloud(DATA$state, DATA$sex, title="Houghton Colors",
max.word.size = 8, min.word.size = .01, X ="purple" , Y = "yellow")
Gradient Cloud
gradient_cloud(DATA2$state, DATA2$sex, title="Gender Word Use", max.word.size = 8, min.word.size = .01, X ="red" , Y = "blue")
Turn of Talk Plot
r VS(1)
tot_plot(mraja1, "dialogue", grouping.var = c("sex", "fam.aff"), tot=FALSE, plot=FALSE)+ scale_fill_brewer(palette="Set1") + geom_hline(aes(yintercept=mean(word.count))) + geom_hline(aes(yintercept=mean(word.count) + (2 *sd(word.count)))) + geom_hline(aes(yintercept=mean(word.count) + (3 *sd(word.count)))) + geom_text(parse=TRUE, hjust=0, vjust=0, size = 3, aes(x = 2, y = mean(word.count) + 2, label = "bar(x)")) + geom_text(hjust=0, vjust=0, size = 3, aes(x = 1, y = mean(word.count) + (2 *sd(word.count)) + 2, label = "+2 sd")) + geom_text(hjust=0, vjust=0, size = 3, aes(x = 1, y = mean(word.count) + (3 *sd(word.count)) + 2, label = "+3 sd")) + ggtitle("Romeo & Juliet:\nAct 1 Turns of Talk")
Venn Diagram
with(DATA , trans_venn(state, person, legend.location = "topright"))
Word Network Plot
word_network_plot(text.var=DATA$state, DATA$person, stopwords=NULL)
Discussion
- How might qdap + R fit into your workflow?
- What do you want to know more about?
- Are there any points that need to be clarified?
- …
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.