Visual Analytics for Linguistics - Day 4 ESSLLI - structured data

Visual Analyticsfor Linguistics -

Olga Scrivner

Course Info

tm Package

StatisticalVisualization

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Visual Analytics for Linguistics - Day 4

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

What You Will Learn

DAY 1 Introduction to Visual Analytics

DAY 2 Visualization Methods, Design, and Tools

DAY 3 Working with Unstructured Data

DAY 4 Working with Structured Data

DAY 5 Advanced Analytics

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Our Materials - Web Site

http://obscrivn.wixsite.com/visualization

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

What We Need

I Interactive Text Mining Suite

I Language Variation Suite

I Download R code

I Download files from Day 3 and Day 4

I R and Rstudio

I R libraries: tm, party, partykit

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

What We Need

I Interactive Text Mining Suite

I Language Variation Suite

I Download R code

I Download files from Day 3 and Day 4

I R and Rstudio

I R libraries: tm, party, partykit

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Text Mining Review

I Intro to ITMS - slides Day 3I R coding

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Text Mining Review

1. Open RStudio2. Open R file textmining.R:

3. Set up working directory

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Text Mining Package - tm

I Main structure - corpus

I Corpus is constructed via DirSource, VectorSource,DataframeSource

mycorpus <- Corpus(VectorSource(file.txt))

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Preprocessing with tm

I lower case

mycorpus <- tm_map(mycorpus, tolower)

I remove punctuation

mycorpus <- tm_map(mycorpus,removePunctuation)

I remove numbers

mycorpus <- tm_map(mycorpus, removeNumbers)

I remove stopwords

mycorpus <- tm_map(mycorpus, removeWords,stopwords(’english’))

I mycorpus <- tm_map(mycorpus, stripWhitespace)

I mycorpus <- tm_map(mycorpus,PlainTextDocument)

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

What is LVS?

Language Variation Suite

It is a Shiny web application originally designed for dataanalysis in sociolinguistic research.

It can be used for:I Processing spreadsheet data

I Visualizing data

I Analyzing means, regression, conditional trees ...(and much more)

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Background

LVS is built in R using Shiny package:

1. R - a free programming language for statisticalcomputing and graphics

2. Shiny App - a web application framework for R

Computational power of R + Web interactivity

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Background

http://littleactuary.github.io/blog/Web-application-framework-with-Shiny/

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Workspace

Browser

I Chrome, Firefox, Safari - recommendable

I Explorer may cause instability issues

AccessibilityI PC, Mac, Linux

I Data files will be uploaded from any location on yourcomputer

I Smart PhoneI Data files must be on a cloud platform connected to

your phone account (e.g. dropbox)

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Server

Since LVS is hosted on a server, Shiny idle time-out settingsmay stop application when it is left inactive (it will grey out).

Solution: Click reload and re-upload your csv file

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Data Preparation

Important things to consider before data entry:I File format:

I Comma separated value (CSV) - faster processingI Excel format will slow processing

I Column names should not contain spacesI Permitted: non-accented characters, numbers,

underscore, hyphen, and period

I One column must contain your dependent variableI The rest of the columns contain independent variables

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Terminology Review

a. Categorical - non-numerical data with two values

I yes - no; male - female

b. Continuous - numerical data

I duration, age, year

c. Multinomial - non-numerical data with three or morevalues

I regions, nationalities

d. Ordinal - scale: currently not supported

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Terminology Review

a. Categorical - non-numerical data with two values

I yes - no; male - female

b. Continuous - numerical data

I duration, age, year

c. Multinomial - non-numerical data with three or morevalues

I regions, nationalities

d. Ordinal - scale: currently not supported

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Workshop Files

1. http://obscrivn.wixsite.com/visualizationDownload file Day 4: movie_metadata.csvSimplified set from https://www.kaggle.com/deepmatrix/imdb-5000-movie-dataset

2. LVS web site: http://languagevariationsuite.com/

http://cl.indiana.edu/~obscrivn/docs/movie_metadata.csv

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Movie Data

I BudgetI DirectorI Actor 1I Director facebook likesI Actor 1 facebook likesI GenreI Year

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Language Variation Suite - Structure

1. Data

I Upload file, data summary, adjust data, cross tabulation

2. Visual Analysis

I Plotting, cluster classification

3. Inferential Statistics

I Modeling, regression, conditional trees, random forest

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

1. Data

2. Visual Analysis

3. Inferential Statistics

I Modeling, regression, conditional trees, random forest

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Upload File

Upload movie_metadata.csv

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Uploaded Dataset

The data content is imported as a table and allows forsorting columns.

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Summary

Summary provides a quantitative summary for each variable,e.g. frequency count, mean, median.

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Data Structure

1. Total number of observations (rows)

2. Number of variables (columns)

3. Variable types

I Factor - categorical valuesI Num - numeric values (0.95, 1.05)I Int - integer values (1, 2, 3)

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Cross Tabulation

Cross-tabulation examines the relationship between variables.

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Cross Tabulation Plot

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Cross Tabulation Plot

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Adjusting Browser - Layout

Shiny pages are fluid and reactive.

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Adjusting Browser - Layout

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

1. Data

2. Visual Analysis

3. Inferential statistics

I Modeling, regression, varbrul analysis, conditional trees,random forest

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

One Variable Plot

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

One Variable Plot

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Customizing Plot

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Customizing Plot

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Saving Plot

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Cluster Plot

I Classification of data into sub-groups is based onpairwise similarities

I Groups are clustered in the form of a tree-likedendrogram

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Cluster Plot

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Cluster Plot

Group 1 Animation, Biography and Group 2 Action, Drama,Comedy

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Inferential Statistics

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

1. Data

2. Visual Analysis

3. Inferential statistics

I Modeling, regression, varbrul analysis, conditional trees,random forest

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

How to Create a Regression Model

Step 1 Modeling - create a model with dependentand independent variables

Step 2 Regression - specify the type of regression(fixed, mixed) and type of dependent variable(binary, continuous, multinomial)

Step 3 Stepwise Regression - compare models(Log-likelihood, AIC, BIC)

Step 4 Conditional Trees - apply non-parametrictests to the model

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Modeling

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Regression Types

I Model

a.) Fixed effect

b.) Mixed effect - individual speaker/token variation (withingroup)

I Type of Dependent Variable

a.) Binary/categorical (only two values)

b.) Continuous (numeric)

c.) Multinomial - categorical with more than two values

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Regression

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Model Output

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Model Output

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Interpretation: Budget and Genre

I Genre Action is the reference value

I Positive coefficient - positive effect

I Negative coefficient - negative effect

http://www.free-online-calculator-use.com/scientific-notation-converter.html

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Interpretation: Budget and Genre

I Genre Action is the reference value

I Positive coefficient - positive effect

I Negative coefficient - negative effecthttp://www.free-online-calculator-use.com/scientific-notation-converter.html

exponential notation:1.46e-7

0.000000146

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Conditional Tree

Conditional tree: a simple non-parametric regression analysis,commonly used in social and psychological studies

I Linear regression: all information is combined linearly

I Conditional tree regression: visual splitting to captureinteraction between variables

Recursive splitting (tree branches)

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Conditional Tree

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Conditional Tree

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Conditional Tree

1. Genre is the significant factor for budget2. Budget distribution is split in two groups:

I Action and Animation

I Biography, Comedy and Drama

3. Budget is significantly higher for Animation and Action

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Random Forest

1. Variable importance for predictors

2. Robust technique with small n large p data

3. All predictors considered jointly (allows for inclusion ofcorrelated factors)

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Random Forest

Let’s add more factors!

I Return to Modeling

I Add independent factors: director facebook likes,actor 1 facebook likes, title year

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Random Forest

I Genre is the most important predictor for this model.I Close to zero or red-dotted line (cut off values) -

irrelevant for this model

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Random Forest

I Genre is the most important predictor for this model.I Close to zero or red-dotted line (cut off values) -

irrelevant for this model

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Assignment

I Select LVS or ITMS

I Upload your own file (csv for LVS or txt/pdf for ITMS)

I Explore your data

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

References I

Baayen, Harald. 2008. Analyzing linguistic data: A practical introduction to statistics.Cambridge: Cambridge University Press

Gries, Stefan Th. 2015. Quantitative designs and statistical techniques. In DouglasBiber Randi Reppen (eds.), The Cambridge Handbook of English Corpus Linguistics.Cambridge: Cambridge University Press

Schnapp, Jeffrey, and Peter Presner. 2009. Digital Humanities Manifesto 2.0.

http://gifsanimados.espaciolatino.com/x_bob_esponja_8.gif

https://daniellestolt.files.wordpress.com/2013/01/are-you-ready1.gif

http://www.martijnwieling.nl/R/sheets.pdf

Olga Scrivner

Course Info

tm Package

Data Preparation

Working withData

Visual Analytics

InferentialAnalysis

Resources

http://www.rdatamining.com/examples/text-mining

https://en.wikibooks.org/wiki/R_Programming/Text_Processing

http://data.library.virginia.edu/reading-pdf-files-into-r-for-text-mining/

http://www.katrinerk.com/courses/words-in-a-haystack-an-introductory-statistics-course/schedule-words-in-a-haystack/r-code-the-text-mining-package

tm package

Visual Analytics for Linguistics - Day 4 ESSLLI - structured data

Data & Analytics

BBI3210 DR AFIDA MOHAMAD ALI. In linguistics, corpus (plural corpora) is a large and structured set of texts (now usually electronically stored,processed

Visualization Design, Methods and Tools - ESSLLI 2017 Workshop Day 2

Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit

Linguistics 101; Learning Basic Linguistics

Visual Analytics for Linguistics - Day 3 ESSLLI

Problems of syntax-semantics interface ESSLLI 02 Trento

Computational Models of Eventsjamespusto.com/.../ESSLLI-2018-Slides-Lecture1-a.pdf · linguistics, logic, AI, robotics, computational linguistics Constituents of events: frame structure,

ESSLLI 2013 Schedule WEEK 1 SLOT 1 WEEK 1 … · ESSLLI 2013 Schedule WEEK 1 ‐ SLOT 1 ... ESSLLI 2013 Schedule WEEK 2 ... Sandra Kübler and Markus Dickinson Extrinsic Parse Improvement

ESSLLI 2002 Workshop E Machine Learning Approaches in G I

Structured programming 4 Day 34 LING 681.02 Computational Linguistics Harry Howard Tulane University

Event Structure with Subevents (Pustejovsky 1991)jamespusto.com/.../ESSLLI-2018-Slides-Lecture1-b.pdf · linguistics, logic, AI, robotics, computational linguistics Constituents of

Proceedings of the ESSLLI Workshop on Distributional ...events.illc.uva.nl/ESSLLI2008/Materials/BaroniEvertLenci/ESSLLI... · ESSLLI Workshop on Distributional Lexical Semantics Bridging

ESSLLI Lecture 4 Normchange - pigozzi.org › ESSLLI Lecture 4 Normchange.pdf · MOTIVATION • Norm change is an interesting open problem. • There are various models of norm change

From Applied Linguistics to Linguistics Applied: Issues ... · British Association for Applied Linguistics From Applied Linguistics to Linguistics Applied: Issues, Practices, Trends

ESSLLI 2006 Summer school Malaga, Spain 31 July – 11 August

ESSLLI 2010: Resource-light Morpho-syntactic Analysis …feldmana/esslli10/1.pdf · Well-structured lists of morphological forms of Sumerian words ... grammar as consisting primarily

ESSLLI Summerschool 2014: Intro to Compositional …user.uni-frankfurt.de/~tezimmer/Zimmermann/ESSLLI-Tuebingen.pdf · Overview Ambiguity: Examples Structural Ambiguity About Meaning

Applied Linguistics Overview of course’ Linguistics’

ESSLLI 2009 Proceedings

An Introduction to Description Logics - esslli2018esslli2018.folli.info/wp-content/uploads/ESSLLI-Part1.pdfIvan Varzinczak (CRIL) An Introduction to Description Logics (Part 1) ESSLLI