A Comparative Review of the R-Instat GUI for R

by Robert A. Muenchen

Introduction

R-Instat is a free and open source graphical user interface for the R software that focuses on people who want to point-and-click their way through data science analyses. Written in Visual Basic, it is currently only available for Microsoft Windows. However, a Linux version is in development using the cross-platform Mono implementation of the .NET framework.This post is one of a series of reviews that aim to help non-programmers choose the Graphical User Interface (GUI) that is best for them. Although I wrote the BlueSky User’s Guide, I hope to remain objective in these reviews. There is no one perfect user interface for everyone; each GUI for R has features that appeal to a different set of people.

Terminology

There are various definitions of user interface types, so here’s how I’ll be using these terms:GUI = Graphical User Interface using menus and dialog boxes to avoid having to type programming code. I do not include any assistance for programming in this definition. So, GUI users are people who prefer using a GUI to perform their analyses. They don’t have the time or inclination to become good programmers.

IDE = Integrated Development Environment which helps programmers write code. I do not include point-and-click style menus and dialog boxes when using this term. IDE users are people who prefer to write R code to perform their analyses.

Installation

The various user interfaces available for R differ quite a lot in how they’re installed. Some, such as jamovi or RKWard, install in a single step. Others, such as Deducer, install in multiple steps (up to seven steps, depending on your needs). Advanced computer users often don’t appreciate how lost beginners can become while attempting even a simple installation. The HelpDesks at most universities are flooded with such calls at the beginning of each semester!

R-Instat is easy to install, requiring only a single step. It provides its own embedded copy of R. This simplifies the installation and ensures complete compatibility between R-Instat and the version of R it’s using. However, it also means if you already have R installed, you’ll end up with a second copy. You can have R-Instat control any version of R you choose, but if the version differs too much, you may run into occasional problems.

Plug-in Modules

When choosing a GUI, one of the most fundamental questions is: what can it do for you? What the initial software installation of each GUI gets you is covered in the Graphics, Analysis, and Modeling sections of this series of articles. Regardless of what comes built-in, it’s good to know how active the development community is. They contribute “plug-ins” that add new menus and dialog boxes to the GUI. This level of activity ranges from very low (RKWard, Rattle, Deducer) through medium (JASP 15) to high (jamovi 43, R Commander 43).

While the R-Instat project welcomes contributions from anyone, there are not any modules to add at this time. All of its capabilities are included in its initial installation.

Startup

Some user interfaces for R, such as jamovi or JASP, start by double-clicking on a single icon, which is great for people who prefer to not write code. Others, such as R commander and JGR, have you start R, then load a package from your library, and then finally call a function. That’s better for people looking to learn R, as those are among the first tasks they’ll have to learn anyway.

You start R-Instat directly by double-clicking its icon from your desktop or choosing it from your Start Menu (i.e., not from within R).

Data Editor

A data editor is a fundamental feature in data analysis software. It puts you in touch with your data and lets you get a feel for it, if only in a rough way. A data editor is such a simple concept that you might think there would be hardly any differences in how they work in different GUIs. While there are technical differences, to a beginner what matters the most are the differences in simplicity. Some GUIs, including jamovi, let you create only what R calls a data frame. They use more common terminology and call it a data set: you create one, you save one, later you open one, then you use one. Others, such as RKWard trade this simplicity for the full R language perspective: a data set is stored in a workspace. So the process goes: you create a data set, you save a workspace, you open a workspace, and choose a data set from within it.

R-Instat starts up by showing its screen (Fig. 1). Under Start, I chose “New Data Frame” and it showed me the rather perplexing dialog shown in Fig. 2.

Figure 1. The R-Instat startup screen.

As an R user, I know what expressions are, but what did the R-Instat designers mean by the term?

Figure 2. The New Dataframe dialog.

Clicking the “Construct Examples” button brought up the suggestions shown in Fig. 3. These are standard R expressions, which came as quite a surprise! It seems that the R-Instat designers are wanting to get people to start using R programming code immediately.

Figure 3. Examples R-Instat provides for expression you can use to create a dataset.

Clicking the Help button brings up the advice, “the simplest option is Empty” (the developers say this will become the default in a future version). Clicking that button brings up a simple prompt for the number of rows and columns you would like to create. After that, you’re looking at a basic spreadsheet (Fig. 4) that easily lets you enter data. As you enter data, it determines if it is numeric or character. Scientific notation is accepted, but dates are saved as character variables. Logical values (TRUE, FALSE) are recognized as such and are stored appropriately.

Right-clicking on any column allows you to convert variables to be a factor, ordered factor, numeric, logical, or character. These changes are recorded as function calls to a custom “convert_column_to_type” function for reproducibility. Such interactive changes are not usually recorded by other R GUIs. Date/time conversion is not available on that menu, as that process is trickier. Those conversions are on the “Prepare> Column Date” menu item. Other things you can do from the right-click menu are: rename, duplicate, reorder, set levels/labels, sort, and filter/remove filter.

The class of each variable is indicated by a character code that follows each variable name in parenthesis: (C) for character, (F) for factor, (O.F) for ordered factor, (D) for date, (L) for logical. When no code follows a variable name, it is numeric.

Figure 4. The R-Instat Data View (left) and Output Window (right).

The name of the dataset appears on a tab at the bottom of the Data View window. This lets you easily manage multiple datasets, an ability that is popular among professionals, but which is rarely offered in R GUIs (BlueSky and R Commander are the only others that offer it).

Once the dataset is saved, to add rows or columns you choose, “Prepare > Data Frame > Insert rows/columns” to add new rows or columns at any position in the data frame. New columns can be added with a specified default value, which can be a big time-saver when entering blocks of related data.

There is a quicker method that works for inserting new rows. You right-click the row numbers and a pop-up menu will allow you to insert rows above or below, and the number of rows selected is the number of rows added – like in Excel.

When editing data, R-Instat lets you type new values on top of the old. As soon as you press the Enter key, it generates R code to execute the change. For example, in a language variable, when changing the value “English” to “Spanish,” it wrote,

Replace Value in Data
data_book$replace_value_in_data(data_name="wakefield", col_name="Language", rows="78", new_value="Spanish")

This is important for reproducibility, but R-Instat is the only GUI reviewed here that tracks such important manual changes. In fact, even among expensive proprietary software, Stata is the only one that I’m aware of that keeps track of such changes using code.

If you have another data set to enter, you can restart the process by choosing “File> New Data…” again. You can change data sets simply by clicking on its tab, and its window will pop to the front for you to see. When doing analyses, or saving data, the data set that is displayed in the editor does not influence what appears in dialog boxes. That means that you can be looking at one dataset while analyzing another! Since each dialog allows you to choose the dataset to use, that is technically not a problem, but if you have several datasets that contain the same variable names, remember that what you see may not be what you get! That’s the opposite of BlueSky Statistics, which automatically analyzes the dataset you see. R-Instat’s ability to work with multiple datasets in a single instance of the software is not a feature found in all R GUIs. For example, jamovi and JASP can only work with a single dataset at a time.

Saving the data is done with a fairly standard “File> Save As> Save Dataset As” menu. By default it will save all open datasets, filters, graphs, and models to a single file called a “data book.” That makes working with complex projects much easier to open and close.

Data Import

R-Instat supports the following file formats, most of which are automatically opened using “File> Import from File”. The ODK and NetCDF file formats have their own Import menus. R-Instat’s ability to open many formats related to climate science hints at what the software excels at. For details, see the Analysis Methods section below.

  1. Comma Separated Values (.csv)
  2. Plain text files (.txt)
  3. Excel (old and new xls file types)
  4. xBASE database files (dBase, etc.)
  5. SPSS (.sav)
  6. SAS binary files (sas7bdat and *.xpt)
  7. Standard R workspace files (RData, but it just opens one dataframe of its choosing)
  8. Open Data Kit (ODK)
  9. OpenRefine
  10. Network Common Data Form (NetCDF)
  11. SST Sea Surface Temperature formatted files
  12. IRI Data Library (API download)
  13. Climate Data Store (CDS) (API download)
  14. Shapefile
  15. Climsoft (Climatic database)
  16. .dly (ASCII files)
  17. .dat (ASCII files)
  18. Tab Separated Values (.tsv)
  19. Stata (.dta)
  20. JSON (.json)
  21. epiinfo (.rec)
  22. Minitab (.mtb)
  23. Systat (.syd). 
  24. CSV with a YAML metadata header (.csvy)
  25. Feather R/Python interchange format (.feather)
  26. Pipe separated files (.psv)
  27. YAML (.yml)
  28. Weka Attribute-Relation File Format (.arff)
  29. Data Interchange Format (.dif)
  30. OpenDocument Spreadsheet (*.ods)
  31. Shallow XML documents (*.xml)
  32. Single-table HTML documents (*.html)

Continued…

BlueSky Statistics Intro and User Guides Now Available

BlueSky Statistics is an easy-to-use menu system that uses the R language to do all its work. My detailed review of BlueSky is available here, and a brief comparison of the various menu systems for R is here. I’ve just released the BlueSky Statistics 7.1 User Guide in printed form on the world’s largest independent bookstore, Lulu.com. A description and detailed table of contents are available here.

Cover design by Kiran Rafiq.

I’ve also released the BlueSky Statistics 7.1 Intro Guide. It is a complete subset of the User Guide, and you can download it for free here (if you have trouble downloading it, your company may have security blocking Microsoft OneDrive; try it at home). Its description and table of contents are here, and soon you will also be able to purchase a printed copy of it from Lulu.com.

Cover design by Kiran Rafiq.

I’m enthusiastic about getting feedback on these books. If you have comments or suggestions, please send them to me at muenchen.bob at gmail dot com.

Other books that feature BlueSky Statistics include:
Introduction to Biomedical Data Science
Applying the Rasch Model in Social Sciences Using R
Data Preparation and Exploration, Applied to Healthcare Data

Publishing with Lulu.com has been a very pleasant experience. They put the author in complete control, making one responsible for every detail of the contents, obtaining reviewers, creating a cover file that includes the front, back, and spine of the book to match the dimensions of the book (e.g. more pages means wider spine, etc.) Advertising is left up to the writer as well, hence this blog post! If you are thinking about writing a book, I highly recommend both Lulu.com and getting a cover design from 99designs.com. The latter let me run a contest in which a dozen artists submitted several ideas each. Their built-in survey system let me ask many colleagues for their opinions to help me decide. Altogether, it was a very interesting experience.

To follow the progress of these and other R related books, subscribe to my blog, or follow me on Twitter.

R GUI Update: BlueSky User’s Guide, New Features

The BlueSky Statistics graphical user interface (GUI) for the R language has added quite a few new features (described below). I’m also working on a BlueSky User Guide, a draft of which you can read about and download here. [Update: don’t download that, get the full Intro Guide download instead.] Although I’m spending a lot of time on BlueSky, I still plan to be as obsessive as ever about reviewing all (or nearly all) of the R GUIs, which is summarized here.

The new data management features in BlueSky are:

  • Date Order Check — this lets you quickly check across the dates stored in many variables, and it reports if it finds any rows whose dates are not always increasing from left to right.
  • Find Duplicates – generates a report of duplicates and saves a copy of the data set from which the duplicates are removed. Duplicates can be based on all variables, or a set of just ID variables.
  • Select First/Last Observation per Group – finding the first or last observation in a group can create new datasets from the “best” or “worst” case in each group, find the most current record, and so on.

Model Fitting / Tuning

One of the more interesting features in BlueSky is its offering of what they call Model Fitting and Model Tuning. Model Fitting gives you direct control over the R function that does the work. That provides precise control over every setting, and it can teach you the code that the menus create, but it also means that model tuning is up to you to do. However, it does standardize scoring so that you do not have to keep up with the wide range of parameters that each of those functions need for scoring. Model Tuning controls models through the caret package, which lets you do things like K-fold cross-validation and model tuning. However, it does not allow control over every model setting.

New Model Fitting menu items are:

  • Cox Proportional Hazards Model: Cox Single Model
  • Cox Multiple Models
  • Cox with Formula
  • Cox Stratified Model
  • Extreme Gradient Boosting
  • KNN
  • Mixed Models
  • Neural Nets: Multi-layer Perceptron
  • NeuralNets (i.e. the package of that name)
  • Quantile Regression

There are so many Model Tuning entries that it’s easier to just paste in the list I updated on the main BlueSkly review that I updated earlier this morning:

  • Model Tuning: Adaboost Classification Trees
  • Model Tuning: Bagged Logic Regression
  • Model Tuning: Bayesian Ridge Regression
  • Model Tuning: Boosted trees: gbm
  • Model Tuning: Boosted trees: xgbtree
  • Model Tuning: Boosted trees: C5.0
  • Model Tuning: Bootstrap Resample
  • Model Tuning: Decision trees: C5.0tree
  • Model Tuning: Decision trees: ctree
  • Model Tuning: Decision trees: rpart (CART)
  • Model Tuning: K-fold Cross-Validation
  • Model Tuning: K Nearest Neighbors
  • Model Tuning: Leave One Out Cross-Validation
  • Model Tuning: Linear Regression: lm
  • Model Tuning: Linear Regression: lmStepAIC
  • Model Tuning: Logistic Regression: glm
  • Model Tuning: Logistic Regression: glmnet
  • Model Tuning: Multi-variate Adaptive Regression Splines (MARS via earth package)
  • Model Tuning: Naive Bayes
  • Model Tuning: Neural Network: nnet
  • Model Tuning: Neural Network: neuralnet
  • Model Tuning: Neural Network: dnn (Deep Neural Net)
  • Model Tuning: Neural Network: rbf
  • Model Tuning: Neural Network: mlp
  • Model Tuning: Random Forest: rf
  • Model Tuning: Random Forest: cforest (uses ctree algorithm)
  • Model Tuning: Random Forest: ranger
  • Model Tuning: Repeated K-fold Cross-Validation
  • Model Tuning: Robust Linear Regression: rlm
  • Model Tuning: Support Vector Machines: svmLinear
  • Model Tuning: Support Vector Machines: svmRadial
  • Model Tuning: Support Vector Machines: svmPoly

You can download the free open-source version from https://BlueSkyStatistics.com.

Updates to R GUIs: BlueSky, jamovi, JASP, & RKWard

Graphical User Interfaces (GUIs) for the R language help beginners get started learning R, help non-programmers get their work done, and help teams of programmers and non-programmers work together by turning code into menus and dialog boxes. There has been quite a lot of progress on R GUIs since my last post on this topic. Below I describe some of the features added to several R GUIs.

BlueSky Statistics

BlueSky Statistics has added mixed-effects linear models. Its dialog shows an improved model builder that will be rolled out to the other modeling dialogs in future releases. Other new statistical methods include quantile regression, survival analysis using both Kaplan-Meier and Cox Proportional Hazards models, Bland-Altman plots, Cohen’s Kappa, Intraclass Correlation, odds ratios and relative risk for M by 2 tables, and sixteen diagnostic measures such as sensitivity, specificity, PPV, NPV, Youden’s Index, and the like. The ability to create complex tables of statistics was added via the powerful arsenal package. Some examples of the types of tables you can create with it are shown here.

Several new dialogs have been added to the Data menu. The Compute Dummy Variables dialog creates dummy (aka indicator) variables from factors for use in modeling. That approach offers greater control over how the dummies are created than you would have when including factors directly in models.

A new Factor Levels menu item leads to many of the functions from the forcats package. They allow you to reorder factor levels by count, by occurrence in the dataset, by functions of another variable, allow you to lump low-frequency levels into a single “Other” category, and so on. These are all helpful in setting the order and nature of, for example, bars in a plot or entries in a table.

The BlueSky Data Grid now has icons that show the type of variable i.e. factor, ordered factor, string, numeric, date or logical. The Output Viewer adds icons to let you add notes to the output (not full R Markdown yet), and a trash can icon lets you delete blocks of output.

A comprehensive list of the changes to this release is located here and my updated review of it is here.

jamovi

New modules expand jamovi’s capabilities to include time-based survival analysis, Bland-Altman analysis & plots, behavioral change analysis, advanced mediation analysis, differential item analysis, and quantiles & probabilities from various continuous distributions.

jamovi’s new Flexplot module greatly expands the types of graphs it can create, letting you take a single graph type and repeat it in rows and/or columns making it easy to visualize how the data is changing across groups (called facet, panel, or lattice plots).

You can read more about Flexplot here, and my recently-updated review of jamovi is here.

JASP

The JASP package has added two major modules, machine learning, and network analysis. The machine learning module includes boosting, K-nearest neighbors, and random forests for both regression and classification problems. For regression, it also adds regularized linear regression. For clustering, it covers hierarchical, K-means, random forest, density-based, and fuzzy C-means methods. It can generate models and add predictions to your dataset, but it still cannot save models for future use. The main method it is missing is a single decision tree model. While less accurate predictors, a simple tree model can often provide insight that is lacking from other methods.

Another major addition to JASP is Network Analysis. It helps you to study the strengths of interactions among people, cell phones, etc. With so many people working from home during the Coronavirus pandemic, it would be interesting to see what this would reveal about how our patterns of working together have changed.

A really useful feature in JASP is its Data Library. It greatly speeds your ability to try out a new feature by offering a completely worked-out example including data. When trying out the network analysis feature, all I had to do was open the prepared example to see what type of data it would use. With most other data science software, you’re left to dig about in a collection of datasets looking for a good one to test a particular analysis. Nicely done!

I’ve updated my full review of JASP, which you can read here.

RKWard

The main improvement to the RKWard GUI for R is adding support for R Markdown. That makes it the second GUI to support R Markdown after R Commander. Both the jamovi and BlueSky teams are headed that way. RKWard’s new live preview feature lets you see text, graphics, and markdown as you work. A comprehensive list of new features is available here, and my full review of it is here.

Conclusion

R GUIs are gaining features at a rapid pace, quickly closing in on the capabilities of commercial data science packages such as SAS, SPSS, and Stata. I encourage R GUI users to contribute their own additions to the menus and dialog boxes of their favorite(s). The development teams are always happy to help with such contributions. To follow the progress of these and other R GUIs, subscribe to my blog, or follow me on twitter.

Biomedical Data Science Textbook Available

By Bob Hoyt & Bob Muenchen

Data science is being used in many ways to improve healthcare and reduce costs. We have written a textbook, Introduction to Biomedical Data Science, to help healthcare professionals understand the topic and to work more effectively with data scientists. The textbook content and data exercises do not require programming skills or higher math. We introduce open source tools such as R and Python, as well as easy-to-use interfaces to them such as BlueSky Statistics, jamovi, R Commander, and Orange. Chapter exercises are based on healthcare data, and supplemental YouTube videos are available in most chapters.

For instructors, we provide PowerPoint slides for each chapter, exercises, quiz questions, and solutions. Instructors can download an electronic copy of the book, the Instructor Manual, and PowerPoints after first registering on the instructor page.

The book is available in print and various electronic formats. Because it is self-published, we plan to update it more rapidly than would be possible through traditional publishers.

Below you will find a detailed table of contents and a list of the textbook authors.

Table of Contents​

​OVERVIEW OF BIOMEDICAL DATA SCIENCE

  1. Introduction
  2. Background and history
  3. Conflicting perspectives
    1. the statistician’s perspective
    2. the machine learner’s perspective
    3. the database administrator’s perspective
    4. the data visualizer’s perspective
  4. Data analytical processes
    1. raw data
    2. data pre-processing
    3. exploratory data analysis (EDA)
    4. predictive modeling approaches
    5. types of models
    6. types of software
  5. Major types of analytics
    1. descriptive analytics
    2. diagnostic analytics
    3. predictive analytics (modeling)
    4. prescriptive analytics
    5. putting it all together
  6. Biomedical data science tools
  7. Biomedical data science education
  8. Biomedical data science careers
  9. Importance of soft skills in data science
  10. Biomedical data science resources
  11. Biomedical data science challenges
  12. Future trends
  13. Conclusion
  14. References

​​SPREADSHEET TOOLS AND TIPS

  1. Introduction
    1. basic spreadsheet functions
    1. download the sample spreadsheet
  2. Navigating the worksheet
  3. Clinical application of spreadsheets
    1. formulas and functions
    2. filter
    3. sorting data
    4. freezing panes
    5. conditional formatting
    6. pivot tables
    7. visualization
    8. data analysis
  4. Tips and tricks
    1. Microsoft Excel shortcuts – windows users
    2. Google sheets tips and tricks
  5. Conclusions
  6. Exercises
  7. References

​​BIOSTATISTICS PRIMER

  1. Introduction
  2. Measures of central tendency & dispersion
    1. the normal and log-normal distributions
  3. Descriptive and inferential statistics
  4. Categorical data analysis
  5. Diagnostic tests
  6. Bayes’ theorem
  7. Types of research studies
    1. observational studies
    2. interventional studies
    3. meta-analysis
    4. orrelation
  8. Linear regression
  9. Comparing two groups
    1. the independent-samples t-test
    2. the wilcoxon-mann-whitney test
  10. Comparing more than two groups
  11. Other types of tests
    1. generalized tests
    2. exact or permutation tests
    3. bootstrap or resampling tests
  12. Stats packages and online calculators
    1. commercial packages
    2. non-commercial or open source packages
    3. online calculators
  13. Challenges
  14. Future trends
  15. Conclusion
  16. Exercises
  17. References

​​DATA VISUALIZATION

  1. Introduction
    1. historical data visualizations
    2. visualization frameworks
  2. Visualization basics
  3. Data visualization software
    1. Microsoft Excel
    2. Google sheets
    3. Tableau
    4. R programming language
    5. other visualization programs
  4. Visualization options
    1. visualizing categorical data
    2. visualizing continuous data
  5. Dashboards
  6. Geographic maps
  7. Challenges
  8. Conclusion
  9. Exercises
  10. References

​​INTRODUCTION TO DATABASES

  1. Introduction
  2. Definitions
  3. A brief history of database models
    1. hierarchical model
    2. network model
    3. relational model
  4. Relational database structure
  5. Clinical data warehouses (CDWs)
  6. Structured query language (SQL)
  7. Learning SQL
  8. Conclusion
  9. Exercises
  10. References

BIG DATA

  1. Introduction
  2. The seven v’s of big data related to health care data
  3. Technical background
  4. Application
  5. Challenges
    1. technical
    2. organizational
    3. legal
    4. translational
  6. Future trends
  7. Conclusion
  8. References

​​BIOINFORMATICS and PRECISION MEDICINE

  1. Introduction
  2. History
  3. Definitions
  4. Biological data analysis – from data to discovery
  5. Biological data types
    1. genomics
    2. transcriptomics
    3. proteomics
    4. bioinformatics data in public repositories
    5. biomedical cancer data portals
  6. Tools for analyzing bioinformatics data
    1. command line tools
    2. web-based tools
  7. Genomic data analysis
  8. Genomic data analysis workflow
    1. variant calling pipeline for whole exome sequencing data
    2. quality check
    3. alignment
    4. variant calling
    5. variant filtering and annotation
    6. downstream analysis
    7. reporting and visualization
  9. Precision medicine – from big data to patient care
  10. Examples of precision medicine
  11. Challenges
  12. Future trends
  13. Useful resources
  14. Conclusion
  15. Exercises
  16. References

​​PROGRAMMING LANGUAGES FOR DATA ANALYSIS

  1. Introduction
  2. History
  3. R language
    1. installing R & rstudio
    2. an example R program
    3. getting help in R
    4. user interfaces for R
    5. R’s default user interface: rgui
    6. Rstudio
    7. menu & dialog guis
    8. some popular R guis
    9. R graphical user interface comparison
    10. R resources
  4. Python language
    1. installing Python
    2. an example Python program
    3. getting help in Python
    4. user interfaces for Python
  5. reproducibility
  6. R vs. Python
  7. Future trends
  8. Conclusion
  9. Exercises
  10. References

​​MACHINE LEARNING

  1. Brief history
  2. Introduction
    1. data refresher
    2. training vs test data
    3. bias and variance
    4. supervised and unsupervised learning
  3. Common machine learning algorithms
  4. Supervised learning
  5. Unsupervised learning
    1. dimensionality reduction
    2. reinforcement learning
    3. semi-supervised learning
  6. Evaluation of predictive analytical performance
    1. classification model evaluation
    2. regression model evaluation
  7. Machine learning software
    1. Weka
    2. Orange
    3. Rapidminer studio
    4. KNIME
    5. Google TensorFlow
    6. honorable mention
    7. summary
  8. Programming languages and machine learning
  9. Machine learning challenges
  10. Machine learning examples
    1. example 1 classification
    2. example 2 regression
    3. example 3 clustering
    4. example 4 association rules
  11. Conclusion
  12. Exercises
  13. References

​​ARTIFICIAL INTELLIGENCE

  1. Introduction
    1. definitions
  2. History
  3. Ai architectures
  4. Deep learning
  5. Image analysis (computer vision)
    1. Radiology
    2. Ophthalmology
    3. Dermatology
    4. Pathology
    5. Cardiology
    6. Neurology
    7. Wearable devices
    8. Image libraries and packages
  6. Natural language processing
    1. NLP libraries and packages
    2. Text mining and medicine
    3. Speech recognition
  7. Electronic health record data and AI
  8. Genomic analysis
  9. AI platforms
    1. deep learning platforms and programs
  10. Artificial intelligence challenges
    1. General
    2. Data issues
    3. Technical
    4. Socio economic and legal
    5. Regulatory
    6. Adverse unintended consequences
    7. Need for more ML and AI education
  11. Future trends
  12. Conclusion
  13. Exercises
  14. References

Authors

Brenda Griffith
Technical Writer
Data.World
Austin, TX

Robert Hoyt MD, FACP, ABPM-CI, FAMIA
Associate Clinical Professor
Department of Internal Medicine
Virginia Commonwealth University
Richmond, VA

David Hurwitz MD, FACP, ABPM-CI
Associate CMIO
Allscripts Healthcare Solutions
Chicago, IL

Madhurima Kaushal MS
Bioinformatics
Washington University at St. Louis, School of Medicine
St. Louis, MO

Robert Leviton MD, MPH, FACEP, ABPM-CI, FAMIA
Assistant Professor
New York Medical College
Department of Emergency Medicine
Valhalla, NY

Karen A. Monsen PhD, RN, FAMIA, FAAN
Professor
School of Nursing
University of Minnesota
Minneapolis, MN

Robert Muenchen MS, PSTAT
Manager, Research Computing Support
University of Tennessee
Knoxville, TN

Dallas Snider PhD
Chair, Department of Information Technology
University of West Florida
Pensacola, FL

​A special thanks to Ann Yoshihashi MD for her help with the publication of this textbook.

SAS Language Clone Now Free for Commercial Use

The WPS Analytics’ version of the SAS language is now available in a Community Edition. This edition allows you to run SAS code on datasets of any size for free. Purchasing a commercial license will get you tech support and the ability to run it from the command line, instead of just interactively. The software license details are listed in this table.

While the WPS version of the SAS language doesn’t do everything the version from SAS Institute offers, it does do quite a lot. The complete list of features is available here.

Back in 2009, the SAS Institute filed a lawsuit against the creators of WPS Analytics, World Programming Limited (WPL), in the High Court of England and Wales. SAS Institute lost the case on the grounds that copyright law applies to software source code, not to its functionality. WPL never had access to SAS Institute’s source code, but they did use a SAS educational license to study how it works. SAS Institute lost another software copyright battle in North Carolina courts, but won over the use of their educational license. SAS Institute is suing a third time, hoping to do better by carefully choosing a pro-patent court in East Texas.

Although I prefer using R, I’m a big fan of the SAS language, as well as SAS Institute, which offers superb technical support. However, I agree with the first two court findings. Copyright law should not apply to a computer language, only to a particular set of source code that creates the language.

Is Scholarly Use of R Beating SPSS Already?

by Bob Muenchen & Sean MacKinnon

One of us (Muenchen) has been tracking The Popularity of Data Science Software using a variety of different approaches. One approach is to use Google Scholar to count the number of scholarly articles found each year for each software. He chose Google Scholar since it searches “across many disciplines and sources: articles, theses, books, abstracts, and court opinions, from academic publishers, professional societies, online repositories, universities, and other web sites.” Figure 1 shows the results from 1995 through 2016. Data collected in 2018 showed that while SPSS use dropped 39% drop from 2017 to 2018, its use was still 66% higher than R in 2018.

Figure 1. Number of citations per year for each statistics package, found by Google Scholar, from 1995 to 2016.

We see in the plot that SPSS was extremely dominant for most of that time period. Even after its precipitous decline, it still beats the rest by more than a 2 to 1 margin. Over the years, several people questioned the accuracy of Figure 1. In a time when scholarly publications are proliferating, how could SPSS use be in such decline?

One hypothesis that has often been suggested revolves around one of the most bizarre product name changes in the history of marketing. As a result of a legal battle for control of the name “SPSS”, the SPSS company changed the name of the product to “PASW”, an acronym for Predictive Analytics Software.  The change made about as much sense as Coke people renaming Coke to “BSW”, for Bubbly Sugar Water. The battle was settled and in 2011 and the product name reverted back to SPSS.

Could that name change account for the apparent decline in its use? A search on Google Scholar from 2009 to 2012 on the string:

“PASW” -“SPSS” -“Amos” 

yielded 12,000 hits. That sounds like quite a few, but when “SPSS” was substituted for “PASW” in that search, we found 701,000 references. At first glance, it seems that the scholarly use of SPSS was undercounted by 1.7%. However, when searching a vast volume of documents, each string may have problems with over-counting. For example, PASW stands for “Plant Available Soil Water” which accounts for 138 of those 12,000 articles. There may be many other such abbreviations. That’s the type of analysis Muenchen did several years ago, before concluding that PASW was more trouble than it was worth (details are here). In 2018 that search yields only 361 hits, and the title of the very first article begins with, “Projections Analysis of Surface Waves (PASW)…”

Muenchen’s hypothesis regarding the apparent decline of SPSS is that it was caused by competition. Back in 2002, SPSS shared the statistical software market with SAS and a couple of others. Its momentum carried it upward for a few more years, then the competition started chipping away at it. GraphPad Prism improved significantly with the release of its version 5 in 2007 and medical users of SPSS found an alternative that was as easy to use while focusing more on their needs.  R added enough useful packages around the same time to become competitive. By now there are probably hundreds of packages that people can use to analyze data, only a few of which are shown in Figure 1.

Mackinnon remained skeptical of this hypothesis because the overall graph appears to show decreases in statistical software citation over time. This would seem to contradict evidence that the number of journal articles published has been increasing at about 3% per year over the last 3 centuries, and about 3.9% per year in the past decade (2018 STM Report, pg. 25). Thus, the total number of citations to statistical software as a collective group should be increasing concurrently with this overall increase.

Mackinnon gathered data from a different source: Scopus. According to Wikipedia, “Scopus covers nearly 36,377 titles from approximately 11,678 publishers, of which 34,346 are peer-reviewed journals in top-level subject fields: life sciences, social sciences, physical sciences, and health sciences.” Mackinnon limited the search to reference lists, reasoning that such citations are likely an indicator of using the software in the paper. Two search strings were used:

REF(“the R software” OR “the R project” OR “r-project.org” OR “R development core”)

REF(SPSS)

These searches are being a bit generous to SPSS by including Modeler and AMOS, and very conservative for R by not including citations to common packages (e.g., ggplot2). The resulting data are plotted in Figure 2.

Figure 2. Number of citations per year for each statistics package, found by Scopus, from 2000 to 2018.

Above we see that the citations of R in scholarly journals exceeded that of SPSS back in 2012. However, the scale of Figure 2 tops out at 30,000 while Figure 1’s scale peaks at 300,000. Google is finding a lot more documents! So, which of these software packages is used the most in scholarly work?  Good question!  We would like to hear your comments below, especially from readers who collect data from other sources.

Data Science Jobs Report 2019: Python Way Up, Tensorflow Growing Rapidly, R Use Double SAS

In my ongoing quest to track The Popularity of Data Science Software, I’ve just updated my analysis of the job market. To save you from reading the entire tome, I’m reproducing that section here.

Job Advertisements

One of the best ways to measure the popularity or market share of software for data science is to count the number of job advertisements that highlight knowledge of each as a requirement. Job ads are rich in information and are backed by money, so they are perhaps the best measure of how popular each software is now. Plots of change in job demand give us a good idea of what is likely to become more popular in the future.

Indeed.com is the biggest job site in the U.S., making its collection of job ads the best around. As their  co-founder and former CEO Paul Forster stated, Indeed.com includes “all the jobs from over 1,000 unique sources, comprising the major job boards – Monster, CareerBuilder, HotJobs, Craigslist – as well as hundreds of newspapers, associations, and company websites.” Indeed.com also has superb search capabilities. It used to have a job trend plotter, but that tool has apparently been shut down.

Searching for jobs using Indeed.com is easy, but searching for software in a way that ensures fair comparisons across packages is challenging. Some software is used only for data science (e.g. SPSS, Apache Spark) while others are used in data science jobs and more broadly in report-writing jobs (e.g. SAS, Tableau). General-purpose languages (e.g. Python, C, Java) are heavily used in data science jobs, but the vast majority of jobs that use them have nothing to do with data science. To level the playing field, I developed a protocol to focus the search for each software within only jobs for data scientists. The details of this protocol are described in a separate article, How to Search for Data Science Jobs. All of the graphs in this section use those procedures to make the required queries.

I collected the job counts discussed in this section on May 27, 2019 and February 24, 2017. One might think that a sample of on a single day might not be very stable, but the large number of job sources makes the counts in Indeed.com’s collection of jobs quite consistent. Data collected in 2017 and 2014 using the same protocol correlated r=.94, p=.002.

Figure 1a shows that Python is in the lead with 27,374 jobs, followed by SQL with  25,877. Java and Amazon’s Machine Learning (ML) tools are roughly 25% further below, with jobs in the 17,000s. R and the C variants come next with around 13,000. People frequently compare R and Python, but when it comes to getting a data science job, there are only half as many for R as for Python. That doesn’t mean they’re the same sort of job, of course. I still see more statisticians using R and machine learning people preferring Python, but Python is definitely on a roll! From Hadoop on down, there is a slow decline in jobs. R is also frequently compared to SAS, which has only 8,123 compared to R’s 13,800.

The scale of Figure 1a is so wide that the bottom package, H20 appears to be zero, when in fact there are 257 jobs for it. 

Figure 1a. Number of data science jobs for the more popular software.

To let us compare the less popular software, I plotted them separately in Figure 1b. Mathematica and Julia are the leaders of this set, with around 219 jobs each. The ancient FORTRAN language is still hanging on to life with 195 jobs. The open source WEKA software and IBM’s Watson are next, with around 185 each. From XGBOOST on down, there is a fairly steady slow decline.

There are several tools that use a workflow interface: Enterprise Miner, KNIME, RapidMiner, and SPSS Modeler. They’re all around the same area between 50 and 100 jobs. In many of the other measures of popularity, RapidMiner beats the very similar KNIME tool, but here there are 50% more jobs for the latter. Alteryx is also a workflow-based tool, however, it has pulled away from the pack, appearing back on Figure 1a with 901 jobs.

Figure 1b. Number of jobs for less popular data science software tools, those with fewer than 250 advertisements.

When interpreting the scale on Figure 1b, what looks like zero is indeed zero. From Systat on down, none of the packages have more than 10 job listings.

It’s important to note that the values shown in Figures 1a and 1b are single points in time. The number of jobs for the more popular software do not change much from day to day. Therefore, the relative rankings of the software shown in Figure 1a is unlikely to change much over the coming year or two. The less popular packages shown in Figure 1b have such low job counts that their ranking is more likely to shift from month to month, though their position relative to the major packages should remain more stable.

Next, let’s look at the change in jobs from the 2017 data to now (2019). Figure 1c shows the percent change for those packages that had at least 100 job listings back in 2017. Without such a limitation, software that goes from 1 job in 2017 to 5 jobs in 2019 would have a 500% increase, but still would be of little interest. Software whose job market is heating up, or growing, is shown in red, while those that are cooling down are shown in blue.

Figure 1c. Percent change in job listings from 2017 to 2019. Only software that had at least 100 jobs in 2017 is shown.

Tensorflow, the deep learning software from Google, is the fastest growing at 523%. Next is Apache Flink, a tool that analyzes streaming data, at 289%. H2O is next, with 150% growth. Caffe is another deep learning framework and its 123% growth reflects the popularity of artificial intelligence algorithms.

Python shows “only” 97% growth, but its popularity was already so high that the 13,471 jobs that it added surpasses the total jobs of many of the other packages!

Tableau is showing a similar rate of growth, though it was a comparably small number of additional jobs, at 4,784.

From the Julia language on down, we see a slowing decrease in growth. I’m surprised to see that jobs for SAS and SPSS are still growing, though barely at 6% and 1%, respectively. 

If you enjoyed reading this article, you might be interested in my recent series of reviews on point-and-click front-ends for the R language. I invite you to subscribe to this blog, or follow me on Twitter.

Comparing Point-and-Click Front Ends for R

For an updated version of this post, see: http://r4stats.com/articles/software-reviews/r-gui-comparison/.

Now that I’ve completed seven detailed reviews of Graphical User Interfaces (GUIs) for R, let’s compare them. It’s easy enough to count their features and plot them, so let’s start there. I’m basing the counts on the number of menu items in each category. That’s not too hard to get, but it’s far from perfect. Some software has fewer menu choices, depending instead on dialog box choices. Studying every menu and dialog box would be too time-consuming, so be aware of this limitation. I’m putting the details of each measure in the appendix so you can adjust the figures and create your own graphs. If you decide to make your own graphs, I’d love to hear from you in the comments below.

Figure 1 shows the number of analytic methods each software supports on the x-axis and the number of graphics methods on the y-axis. The analytic methods count combines statistical features, machine learning / artificial intelligence ones (ML/AI), and the ability to create R model objects. The graphics features count totals up the number of bar charts, scatterplots, etc. each package can create.

The ideal place to be in this graph is in the upper right corner. We see that BlueSky and R Commander offer quite a lot of both analytic and graphical features. Rattle stands out as having the second greatest number of graphics features. JASP is the lowest on graphics features and 3rd from the bottom on analytic ones.

Next, let’s swap out the y-axis for general usability features. These consist of a variety of features that make your work easier, including data management capabilities (see appendix for details).

Figure 2 shows that BlueSky and R Commander still in the top two positions overall, but now Deducer has nearly caught up with R Commander on the number of general features. That’s due to its reasonably strong set of data management tools, plus its output is in true word processing tables saving you the trouble of formatting it yourself. Rattle is much lower in this plot since, while its graphics capabilities are strong (at least in relation to ML/AI tasks), it has minimal data management capabilities.

These plots help show us three main overall feature sets, but each package offers things that the others don’t. Let’s look at a brief overview of each. Remember that each of these has a detailed review that follows my standard template. I’ll start with the two that have come out on top, then follow in alphabetical order.

The R Commander – This is the oldest GUI, having been around since at least 2005. There are an impressive 41 plug-ins developed for it. It is currently the only R GUI that saves R Markdown files, but it does not create word processing tables by default, as some of the others do. The R code it writes is classic, rarely using the newer tidyverse functions. It works as a partner to R; you install R separately, then use it to install and start R Commander. It makes it easy to blend menu-based analysis with coding. If your goal is to learn to code in classic R, this is an excellent choice.

BlueSky Statistics – This software was created by former SPSS employees and it shares many of SPSS’ features. BlueSky is only a few years old, and it converted from commercial to open source just a few months ago. Although BlueSky and R Commander offer many of the same features, they do them in different ways. When using BlueSky, it’s not initially apparent that R is involved at all. Unless you click the “Syntax” button that every dialog box has, you’ll never see the R code or the code editor. Its output is in publication-quality tables which follow the popular style of the American Psychological Association.

Deducer – This has a very nice-looking interface, and it’s probably the first to offer true word processing tables by default. Being able to just cut and paste a table into your word processor saves a lot of time and it’s a feature that has been copied by several others. Deducer was released in 2008, and when I first saw it, I thought it would quickly gain developers. It got a few, but development seems to have halted. Deducer’s installation is quite complex, and it depends on the troublesome Java software. It also used JGR, which never became as popular as the similar RStudio. The main developer, Ian Fellows, has moved on to another very interesting GUI project called Vivid.

jamovi – The developers who form the core of the jamovi project used to be part of the JASP team. Despite the fact that they started a couple of years later, they’re ahead of JASP in several ways at the moment. Its developers decided that the R code it used should be visible and any R code should be executable, something that differentiated it from JASP. jamovi has an extremely interactive interface that shows you the result of every selection in each dialog box. It also saves the settings in every dialog box, and lets you re-use every step on a new dataset by saving a “template.” That’s extremely useful since GUI users often don’t want to learn R code. jamovi’s biggest weakness its dearth of data management tasks, though there are plans to address that.

JASP – The biggest advantage JASP offers is its emphasis on Bayesian analysis. If that’s your preference, this might be the one for you. At the moment JASP is very different from all the other GUIs reviewed here because it won’t show you the R code it’s writing, and you can’t execute your own R code from within it. Plus the software has not been open to outside developers. The development team plans to address those issues, and their deep pockets should give them an edge.

Rattle – If your work involves ML/AI (a.k.a. data mining) instead of standard statistical methods, Rattle may be the best GUI for you. It’s focused on ML/AI, and its tabbed-based interface makes quick work of it. However, it’s the weakest of them all when it comes to statistical analysis. It also lacks many standard data management features. The only other GUI that offers many ML/AI features is BlueSky.

RKWard – This GUI blends a nice point-and-click interface with an integrated development environment that is the most advanced of all the other GUIs reviewed here. It’s easy to install and start, and it saves all your dialog box settings, allowing you to rerun them. However, that’s done step-by-step, not all at once as jamovi’s templates allow. The code RKWard creates is classic R, with no tidyverse at all.

Conclusion

I hope this brief comparison will help you choose the R GUI that is right for you. Each offers unique features that can make life easier for non-programmers. If one catches your eye, don’t forget to read the full review of it here.

Acknowledgements

Writing this set of reviews has been a monumental undertaking. It would not have been possible without the assistance of Bruno Boutin, Anil Dabral, Ian Fellows, John Fox, Thomas Friedrichsmeier, Rachel Ladd, Jonathan Love, Ruben Ortiz, Christina Peterson, Josh Price, Eric-Jan Wagenmakers, and Graham Williams.

Appendix: Guide to Scoring

In figures 1 and 2, Analytic Features adds up: statistics, machine learning / artificial intelligence, the ability to create R model objects, and the ability to validate models using techniques such as k-fold cross-validation. The Graphics Features is the sum of two rows, the number of graphs the software can create plus one point for small multiples, or facets, if it can do them. Usability is everything else, with each row worth 1 point, except where noted.

FeatureDefinition
Simple
installation
Is it done in one step?
Simple start-upDoes it start on its own without starting R, loading
packages, etc.?
Import Data FilesHow many files types can it import?
Import
Database
How many databases can it read from?
Export Data FilesHow many file formats can it write to?
Data EditorDoes it have a data editor?
Can work on >1 fileCan it work on more than one file at a time?
Variable
View
Does it show metadata in a variable view, allowing for many fast edits to metadata?
Data
Management
How many data management tasks can it do?
Transform
Many
Can it transform many variables at once?
Graph TypesHow many graph types does it have?
Small
Multiples
Can it show small multiples (facets)?
Model ObjectsCan it create R model objects?
StatisticsHow many statistical methods does it have?
ML/AIHow many ML / AI methods does it have?
Model ValidationDoes it offer model validation (k-fold, etc.)?
R Code IDECan you edit and execute R code?
GUI ReuseDoes it let you re-use work without code?
Code ReuseDoes it let you rerun all using code?
Package ManagementDoes it manage packages for you?
Table of ContentsDoes output have a table of contents?
Re-orderCan you re-order output?
Publication QualityIs output in publication quality by default?
R MarkdownCan it create R Markdown?
Add
comments
Can you add comments to output?
Group-by Does it do group-by repetition of any other task?
Output as
Input
Does it save equivalent to broom’s tidy, glance, augment? (They earn 1 point for each)
Developer toolsDoes it offer developer tools?

Scores

FeatureBlueSkyDeducerJASPjamoviRattleRcmdrRKWard
Simple installation1011001
Simple start-up1111001
Import Data Files71345975
Import Database5000100
Export Data Files5714133
Data Editor1101011
Can work on >1 file1100000
Variable View1100000
Data Management309239254
Transform Many1101110
Graph Types2516912242114
Small Multiples1100010
Model Objects1100011
Statistics9637264489522
ML/AI90001200
Model Validation1000100
R Code IDE1101001
GUI Reuse0010001
Code Reuse1101111
Package Management1011000
Output: Table of Contents1000000
Output: Re-order0000000
Output: Publication Quality1111000
Output: R Markdown0000010
Output: Add comments0010010
Group-by / Split File1000000
Output as Input3100010
Developer tools1101011
Total1969448776716056

Data Science Software Used in Journals: Stat Packages Declining (including R), AI/ML Software Growing

In my neverending quest to track The Popularity of Data Science Software, it’s time to update the section on Scholarly Articles. The rapid growth of R could not go on forever and, as you’ll see below, its use actually declined over the last year.

Scholarly Articles

Scholarly articles provide a rich source of information about data science tools. Because publishing requires significant amounts of effort, analyzing the type of data science tools used in scholarly articles provides a better picture of their popularity than a simple survey of tool usage. The more popular a software package is, the more likely it will appear in scholarly publications as an analysis tool, or even as an object of study.

Since scholarly articles tend to use cutting-edge methods, the software used in them can be a leading indicator of where the overall market of data science software is headed. Google Scholar offers a way to measure such activity. However, no search of this magnitude is perfect; each will include some irrelevant articles and reject some relevant ones. The details of the search terms I used are complex enough to move to a companion article, How to Search For Data Science Articles.  Since Google regularly improves its search algorithm, each year I collect data again for the previous years (with one exception noted below).

Figure 2a shows the number of articles found for the more popular software packages and languages (those with at least 1,700 articles) in the most recent complete year, 2018. To allow ample time for publication, insertion into online databases, and indexing, the was data collected on 3/28/2019.

Figure 2a. The number of scholarly articles found on Google Scholar, for data science software. Only those with more than 1,700 citations are shown.

SPSS is by far the most dominant package, as it has been for over 20 years. This may be due to its balance between power and ease-of-use. R is in second place with around half as many articles. It offers extreme power, though with less ease of use. SAS is in third place, with a slight lead over Stata, MATLAB, and GraphPad Prism, which are nearly tied.

Note that the general-purpose languages: C, C++, C#, FORTRAN, Java, MATLAB, and Python are included only when found in combination with data science terms, so view those counts as more of an approximation than the rest.

The next group of packages goes from Python through C, with usage declining slowly. The next set starts at Caffe, dropping nearly 50%, and continuing to IBM Watson with a slow decline.

The last two packages in Fig 2a are Weka and Theano, which are quite a drop from IBM Watson, though it’s getting harder to see as the lines shrink.

To continue on this scale would make the remaining packages all appear too close to the y-axis to read, so Figure 2b shows the remaining software on a much smaller scale, with the y-axis going to only 1,700 rather than the 80,000 used on Figure 2a.

Figure 2b. Number of scholarly articles using each data science software found using Google Scholar. Only those with fewer than 1,700 citations are shown.

I chose to begin Figure 2b with software that has fewer than 1,700 articles because it allows us to see RapidMiner and KNIME on the same scale. They are both workflow-driven tools with very similar capabilities. This plot shows RapidMiner with 49% greater usage than KNIME. RapidMiner uses more marketing, while KNIME depends more on word-of-mouth recommendations and a more open source model. The IT advisory firms Gartner and Forrester rate them as tools able to hold their own against the commercial titans, IBM’s SPSS and SAS. Given that SPSS has roughly 50 times the usage in academia, that seems like quite a stretch. However, as we will soon see, usage of these newer packages are growing, while the use of the older ones is shrinking quite rapidly.

Figure 2b also lets us see IBM’s SPSS Modeler, SAS Enterprise Miner, and Alteryx on the same plot. These three are also workflow-driven tools which are quite expensive. None are doing as well here as RapidMiner or KNIME, tools that much less expensive – or free – depending on how you use them (KNIME desktop is free but server is not; RapidMiner is free for analyzing fewer than 10,000 cases).

Another interesting comparison on Figure 2b is JASP and jamovi. Both are open-source tools that focus on statistics rather than machine learning or artificial intelligence. They both use graphical user interfaces (GUIs) in a style that is similar to SPSS. Both also use R behind the scenes to do their calculations. JASP emphasizes Bayesian Analysis and hides its R code; jamovi has a more frequentist orientation, it lets you see its R code, and it lets you execute your own R code directly from within it. JASP currently has nine times as many citations here, though jamovi’s use is growing much more rapidly.

Even newer on the GUI for R scene is BlueSky Statistics, which doesn’t appear on the plot at all since it has zero scholarly articles so far. It was created by a new company and only adopted an open source model a few months ago.

While Figures 2a and 2b are useful for studying market share as it stands now, they don’t show how things are changing. It would be ideal to have long-term growth trend graphs for each of the analytics packages, but collecting that much data annually is too time-consuming. What I’ve done instead is collect data only for the past two complete years, 2017 and 2018. This provides the data needed to study year-over-year changes.

Figure 2c shows the percent change across those years, with the growing “hot” packages shown in red (right side); the declining or “cooling” are shown in blue (left side). Since the number of articles tends to be in the thousands or tens of thousands, I have removed any software that had fewer than 1,000 articles in 2015. A package that grows from 1 article to 5 may demonstrate 500% growth but is still of little interest.

Figure 2c. Change in Google Scholar citation rate in the most recent complete two years, 2017 and 2018.

The recent changes in data science software can be summarized succinctly: AI/ML up; statistics down. The software that is growing contains none of the packages that are associated more with statistical analysis. The software in decline is dominated by the classic packages of statistics: SPSS Statistics, SAS, GraphPad Prism, Stata, Statgraphics, R, Statistica, Systat, and Minitab. JMP is the only traditional statistics package whose scholarly usage is growing. Of the machine learning software that’s declining in usage, there are rough equivalents that are growing (e.g. Mahout down, Spark up).

Of course another summary is: cheap (or free) up; expensive down. Of the growing packages, 13 out of 17 are available in open source. Of those in decline, only 5 out of 13 are open source.

Statistics software has been around much longer than AI/ML software, started back in the days before open source. Stat vendors have been adding AI/ML methods to their software, making them the more comprehensive solutions. The AI/ML vendors or projects are missing an opportunity to add more comprehensive statistics capabilities. Some, such as RapidMiner and KNIME, are indeed expanding in this direction, but very slowly indeed.

At the top of Figure 2c, we see that the deep learning packages Keras and TensorFlow are the fastest growing at nearly 150%. PyTorch is not shown here because it did not have enough usage in the previous year. However, its citation rate went from 616 to 4,670, a substantial 658% growth rate! There are other packages that are not shown here, including JASP with 223% growth, and jamovi with 720% growth. Despite such high growth, the latter still only has 108 citations in 2018. The rapid growth of JASP and jamovi lend credence to the perspective that the overall pattern of change shown in Figure 2c may be more of a result of free vs. expensive software. Neither of them offers any AI/ML features.

Scikit Learn, the Python machine learning library, was a fast grower with a 60% increase.

I was surprised to see IBM Watson growing a healthy 34% as much of the news about it has not been good. It’s awesome at Jeopardy though!

In the RapidMiner vs. KNIME contest, we saw previously that RapidMiner was ahead. From this plot, we that KNIME growing slightly (5.7%) while RapidMiner is declining slightly (1.8%).

The biggest losers in Figure 2c are SPSS, down 39%, and SAS, Prism, and Mahout, all down 24%. Even R is down 13%. Recall that Figure 2a shows that despite recent years of decline, SPSS is still extremely dominant for scholarly use, and R and SAS are still the #2 and #3 most widely used packages in this arena.

I’m particularly interested in the long-term trends of the classic statistics packages. So in Figure 2d I have plotted the same scholarly-use data for 1995 through 2016.

Figure 2d. The number of Google Scholar citations for each classic statistics package per year from 1995 through 2016.

SPSS has a clear lead overall, but now you can see that its dominance peaked in 2009 and its use is in sharp decline. SAS never came close to SPSS’ level of dominance, and its use peaked around 2010. GraphPAD Prism followed a similar pattern, though it peaked a bit later, around 2013.

In Figure 2d, the extreme dominance of SPSS makes it hard to see long-term trends in the other software. To address this problem, I have removed SPSS and all the data from SAS except for 2014 and 1015. The result is shown in Figure 2e.

Figure 2e. The number of Google Scholar citations for each classic statistics package from 1995 through 2016, this time with SPSS removed and SAS included only in 2014 and 2015. The removal of SPSS and SAS expanded scale makes it easier to see the rapid growth of the less popular packages.

Figure 2e makes it easy to see that most of the remaining packages grew steadily across the time period shown. R and Stata grew especially fast, as did Prism until 2012. Note that the decline in the number of articles that used SPSS, SAS, or Prism is not balanced by the increase in the other software shown in this particular graph. Even adding up all the other software shown in Figures 2a and 2b doesn’t account for the overall decline. However, I’m looking at only 58 out of over 100 data science tools.

While Figures 2d and 2e show the historical trend that ended in 2016, Figure 2f shows a fresh set of data collected in March, 2019. Since Google’s algorithm changes, preventing the new data from matching exactly with the old, this new data starts at 2015 so the two sets overlap. SPSS is not shown on this graph because its dominance would compress the y-axis, making trends in the others harder to see. However, keep in mind that despite SPSS’ 39% drop from 2017 to 2018, its use is still 66% higher than R’s in 2018! Apparently people are willing to pay for ease of use.


Figure 2f. The number of Google Scholar citations for each classic statistics package per year from 2015 through 2018.

In Figure 2f we can see that the downward trends of SAS, Prism, and Statistica are continuing. We also see that the long and rapid growth of R and Stata has come to an end. Growth that rapid can’t go on forever. It will be interesting to see next year to see if this is merely a flattening of usage or the beginning of a declining trend. As I pointed out in my book, R for Stata Users, there are many commonalities between R and Stata. As a result of this, and the fact that R is open source, I expect R use to stabilize at this level while use of Stata continues to slowly decline.

SPSS’ long-term rapid decline has to level out at some point. They have been chipped away at by many competitors. However, until recently these competitors have either been free and code-based such as R, or menu-based and proprietary, such as Prism. With the fairly recent arrival of JASP, jamovi, and BlueSky Statistics, SPSS now faces software that is both free and menu-based. Previous projects to add menus to R, such as the R Commander and Deducer, were also free and open source, but they required installing R separately and then using R code to activate the menus.

These results apply to scholarly articles in general. The results in specific fields or journals are very likely to be different.

To see many other ways to estimate the market share of this type of software, see my ongoing article, The Popularity of Data Science Software. My next post will update the job advertisements that list science software. You may also be interested in my in-depth reviews of point-and-click user interfaces to R. I invite you to subscribe to my blog or follow me on twitter where I announce new posts. Happy computing!