R GUI Reviews Updated

I have just finished updating my reviews of graphical user interfaces for the R language. These include BlueSky Statistics, jamovi, JASP, R AnalyticFlow, R Commander, R-Instat, Rattle, and RKward. The permanent link to the article that summarizes it all is https://r4stats.com/articles/software-reviews/r-gui-comparison/.
I list the highlights below as this post to reach all the blog aggregators. If you have suggestions for improving any of the reviews, please let me know at muenchen.bob@gmail.com.

With so many detailed reviews of Graphical User Interfaces (GUIs) for R available, which should you choose? It’s not too difficult to rank them based on the number of features they offer, so I’ll start there. Then, I’ll follow with a brief overview of each.

I’m basing the counts on the number of dialog boxes in each category of the following categories:

  • Ease of Use
  • General Usability
  • Graphics
  • Analytics
  • Reproducibility

This data is trickier to collect than you might think. Some software has fewer menu choices, depending on more detailed dialog boxes instead. Studying every menu and dialog box is very time-consuming, but that is what I’ve tried to do to keep this comparison trustworthy. Each development team has had a chance to look the data over and correct errors.

Perhaps the biggest flaw in this methodology is that every feature adds only one point to each GUI’s total score. I encourage you to download the full dataset to consider which features are most important to you. If you decide to make your own graphs with a different weighting system, I’d love to hear from you in the comments below.

Ease of Use

For ease of use, I’ve defined it primarily by how well each GUI meets its primary goal: avoiding code. They get one point for each of the following abilities, which include being able to install, start, and use the GUI to its maximum effect, including publication-quality output, without knowing anything about the R language itself. Figure one shows the result. R Commander is abbreviated Rcmdr, and R AnalyticFlow is abbreviated RAF. The commercial BlueSky Pro comes out on top by a slim margin, followed closely by JASP and RKWard. None of the GUIs achieved the highest possible score of 14, so there is room for improvement.

  • Installs without the use of R
  • Starts without the use of R
  • Remembers recent files
  • Hides R code by default
  • Use its full capability without using R
  • Data editor included
  • Pub-quality tables w/out R code steps
  • Simple menus that grow as needed
  • Table of Contents to ease navigation
  • Variable labels ease identification in the output
  • Easy to move blocks of output
  • Ease reading columns by freezing headers of long tables
  • Accepts data pasted from the clipboard
  • Easy to move header row of pasted data into the variable name field
Figure 1. The number of ease of use features offered by each R GUI.

General Usability

This category is dominated by data-wrangling capabilities, where data scientists and statisticians spend most of their time. It also includes various types of data input and output. We see in Figure 2 that both BlueSky versions and R-Instat come out on top not just due to their excellent selection of data-wrangling features but also for their use of the rio package for importing and exporting files. The rio package combines the import/export capabilities of many other packages, and it is easy to use. I expect the other GUIs will eventually adopt it, raising their scores by around 20 points.

  • Operating systems (how many)
  • Import data file types (how many)
  • Import from databases (how many)
  • Export data file types (how many)
  • Languages displayable in UI (how many, besides English)
  • Easy to repeat any step by groups (split-file)
  • Multiple data files open at once
  • Multiple output windows
  • Multiple code windows
  • Variable metadata view
  • Variable types (how many)
  • Variable search/filter in dialogs
  • Variable sort by name
  • Variable sort by type
  • Variable move manually
  • Model Builder (how many effect types)
  • Magnify GUI for teaching
  • R code editor
  • Comment/uncomment blocks of code
  • Package management (comes with R and all packages)
  • Output: word processing features
  • Output: R Markdown
  • Output: LaTeX
  • Data wrangling (how many)
  • Transform across many variables at once (e.g., row mean)
  • Transform down many variables at once (e.g., log, sqrt)
  • Assign factor labels across many variables at once
  • Project saves/loads data, dialogs, and notes in one file
Figure 2. The number of general usability features in each R GUI.

Graphics

This category consists mainly of the number of graphics each software offers. However, the other items can be very important to completing your work. They should add more than one point to the graphics score, but I scored them one point since some will view them as very important while others might not need them at all. Be sure to see the full reviews or download the Excel file if those features are important to you. Figure 3 shows the total graphics score for each GUI. R-Instat has a solid lead in this category. In fact, this underestimates R-Instat’s ability if you include its options to layer any “geom” on top of another graph. However, that requires knowing the geoms and how to use them. That’s knowledge of R code, of course.

When studying these graphs, it’s important to consider the difference between the relative and absolute performance. For example, relatively speaking, R Commander is not doing well here, but it does offer over 25 types of plots! That absolute figure might be fine for your needs.

Continued…

BlueSky Statistics Version 10 is Not Open Source

BlueSky Statistics is a graphical user interface for the powerful R language. On July 10, 2024, the BlueskyStatistics.com website said:

“…As the BlueSky Statistics version 10 product evolves, we will continue to work on orchestrating the necessary logistics to make the BlueSky Statistics version 10.x application available as an open-source project. This will be done in phases, as we did for the BlueSky Statistics 7.x version. We are currently rearchitecting its key components to allow the broader community to make effective contributions. When this work is complete, we will open-source the components for broader community participation…”

In the current statement (September 5, 2024), the sentence regarding version 10.x becoming open source is gone. This line was added:

“…Revenue from the commercial (Pro) version plays a vital role in funding the R&D needed to continue to develop and support the open-source (BlueSky Statistics 7.x) version and the free version (BlueSky Statistics 10.x Base Edition)…”

I have verified with the founders that they no longer plan to release version 10 with an open-source license. I’m disappointed by this change as I have advocated for and written about open source for many years.

There are many advantages of open-source licensing over proprietary. If the company decides to stop making version 10 free, current users will still have the right to run the currently installed version, but they will only be able to get the next version if they pay. If it were open source, its users could move the code to another repository and base new versions on that. That scenario has certainly happened before, most notably with OpenOffice. BlueSky LLC has announced no plans to charge for future versions of BlueSky Base Edition, but they could.

I have already updated the references on my website to reflect that BlueSky v10 is not open source. I wish I had been notified of this change before telling many people at the JSM 2024 conference that I was demonstrating open-source software. I apologize to them.

BlueSky Statistics Enhancements

BlueSky Statistics is a free and open-source graphical user interface for the powerful R language. There is also a commercial “Pro” version that offers tech support, priority feature requests, and many powerful additional features. The Pro version has been beefed up considerably with the new features below. These features apply to quality control, general statistics, team collaboration, project management, and scripting. Many are focused on quality control and Six Sigma as a result of requests from organizations migrating from Minitab and JMP. However, both versions of BlueSky Statistics offer a wide range of statistical, graphical, and machine-learning methods.

The free version saves every step of the analysis for full reproducibility. However, repeating the analysis is a step-by-step process. The Pro version can now rerun the entire set at once, substituting other datasets when needed.

You can obtain either version at https://BlueSkyStatistics.com. A detailed review is available at https://r4stats.com/articles/software-reviews/bluesky/. If you plan to attend the Joint Statistical Meetings (JSM) in Portland next week, stop by Booth 406 to get a demonstration. We hope to see you there!

Copy and Paste data from Excel

      Copy data from Excel and paste it into the BlueSky Statistics data grid. This is in addition to the existing mechanism of bringing data through file import for various file formats into BlueSky Statistics to perform data analysis.

      Undo/Redo data grid edits

      Single-item and muti-items data element edits can be discarded by undo and restored by redo operations.

      Project save/open to save/open all work (all open datasets and output analysis)

      Analysis performed can be saved into one or more projects. Each project contains all the datasets along with all the analyses and any R code from the editor. The projects can be exported and shared (sent as .bsp, which is a zip file; “bsp” is an abbreviation of BlueSky Statistics Project) with other BlueSky Statistics users. The users can import projects, see all the datasets and analyses stored in the projects, and subsequently add/modify/rerun all the analyses.

      Enhanced cleaning/adjustment of copied/imported Excel/CSV data on the Datagrid

      Dataset > Excel Cleanup

      There are a few enhancements made to offer additional data cleanup/adjustment options to the existing Excel Cleanup dialog to clean/adjust (i.e., rows. Columns, data type, etc.) data on the BlueSky Statistics data grid, irrespective of how the data was loaded into the data grid with the file open option or by copying and pasting from Excel/CSV file.

      Renaming output tabs

      Double-clicking on the output tab will open a dialog box asking for the new name. The user can type in a name to rename the output tab.

      Enhanced Pie Chart and Bar Chart

      Graphics > Pie Charts > Pie Chart
      Graphics > Bar Chart

                    The pie chart and bar chart have been enhanced to show % and counts on the plot.

      Scatterplot Matrix

      Graphics > Scatterplot Matrix

      The Scatter Plot Matrix dialog has been added.

      Scatterplot with mean and confidence interval bar

      Graphics > Scatterplot > Scatter Plot with Intervals

      A Scatter Plot dialog with mean and confidence interval bar has been made available with an unlimited number of grouping variables for the X-axis to group a numeric variable for the Y-axis.

      Enhanced Scatterplot with both horizontal and vertical reference lines

      Graphics > Scatterplot > Scatter Plot Ref Lines

      The Scatterplot dialog has been enhanced so that users can add an unlimited number of reference lines (horizontal and vertical axis) to the plot. 

      Enhancements to BlueSky Statistics R Editor and Output Syntax/Code Editor

      For R-programmers many enhancements have been made to the BlueSky R Editor and the output syntax/code editor to improve ease of use and productivity with tooltips, find and replace, undo/redo, comment/uncomment blocks, etc.

      Enhanced Normal Distribution Plot

        Distribution > Normal > Normal Distribution Plot with Labels 

        • The normal distribution plot will show the computed probability and x values on the plot for the shaded area for x value and quantiles, respectively
        • Plot one tail (left and right), two tails, and other ranges

        Automatic randomization of generating normal sample distribution

          Distribution > Normal > Sample from Normal Distribution

          In addition to setting a seed value for reproducibility, the default option has been set to randomize automatically the sample data generation every time.

          Automatic randomization of design creations of all DoE designs

            DOE > Create Design > …. 

            In addition to setting a seed value for reproducibility, the default option has been set to randomize the creation of any DoE design every time automatically.

            Enhanced Distribution Fit analysis

              Analysis > Distribution Analysis > Distribution Fit P-value

              The distribution fit analysis has been enhanced to compute AD, KS, and CVM tests and show test statistics, as well as corresponding p-values. These assist users in determining the best fit in addition to the existing AIC and BIC values.

              Moreover, an option has been introduced for users to see only the comparison of distributions and skip displaying the analysis of the individual distribution fit analysis.

              Tolerance Intervals

                Six Sigma > Tolerance Intervals

                A new Tolerance Intervals analysis has been introduced. The tolerance interval describes the range of values for a distribution with confidence limits calculated to a particular percentile of the distribution. These tolerance limits, taken from the estimated interval, are limits within which a stated proportion of the population is expected to occur.

                Equivalence (and Minimal Effect) test

                  Analysis > Means > Equivalence test

                  This new feature tests for mean equivalence and minimal effects.

                  Nonlinear Least Square – all-purpose Non-Linear Regression modeling

                    Model Fitting > Nonlinear Least Square

                    Performs non-linear regression with flexibility and many user options to model, test, and plot.

                    Polynomial Models with different degrees

                      Model Fitting > Polynomial

                      Computes and fits an orthogonal polynomial model with a specified degree. Also, optionally compares multiple Polynomial models of different degrees side by side. 

                      Enhanced Pareto Chart

                        Six Sigma > Pareto Chart > Pareto Chart

                        A new option has been added for data that does not have a count column but only has the raw data. Automatically computes cumulative frequency from Raw Data for plotting.

                        Frequency analysis with an option to draw a Pareto chart

                          Analysis > Summary > Frequency Plot

                          A new dialog has been introduced to plot (optionally) the Pareto Chart from the frequency table and, if desired, display the frequency table on the Datagrid.

                          MSA (Measurement System Analysis) Enhancements

                          Gage Study Design Table

                          Six Sigma > MSA > Design MSA Study

                          Users can generate a randomized design experiment table for any combination of the number of operators, parts, and replications to set up a Gage study table to perform experiments and collect the results to analyze the accuracy of the Gage under study with analysis like Gage R&R, Gage Bias, etc.

                          Enhanced Gage R&R

                          Six Sigma > MSA > Gage R&R

                          Many enhancements and options have been introduced to the Gage of R&R dialog and the underlying analysis

                          • Report header table
                          • Enlarged graphs
                          • Nested gage data analysis, in addition to crossed
                          • Usage of historical process std dev to estimate Gage Evaluation values (%StudyVar table)
                          • Show %Process

                          Enhanced Gage Attribute Analysis

                          Six Sigma > MSA > Attribute Analysis

                          Many enhancements and options have been introduced to the Attribute Analysis dialog and the underlying analysis

                          • Report header table
                          • Accuracy and classification rate calculations, in addition to agreement and disagreement
                          • Optional Cohen’s Kappa stats (between each pair of raters) in addition to Fleiss Kappa (multi-raters)

                          Enhanced Gage Bias Analysis

                          Six Sigma > MSA > Gage Bias Analysis

                          Many enhancements and options have been introduced to the Gage Bias Analysis dialog and the underlying analysis

                          • Efficient single dialog with options for linearity and type-1 tests for one or more References
                          • A new option – “Method to use for estimating repeatability std dev”
                          • Cg and Cgk – calculated for different Reference values in one go
                          • Run charts for every reference value and an overall run chart for all reference values
                          • Usage of historical std dev to calculate RF (Reference Figure)
                          • %RE, %EV are introduced, and all tables show how the computed values compared to the required/cut-off values specified by users on the dialog

                          PCA (Process Capability Analysis) Enhancements

                          Enhanced Process Capability Analysis (for normal data)

                          Six Sigma > Process Capability > Process Capability

                          •  pp_l = pp_k and ppU = ppk is shown when a one-sided tolerance is used
                          • Removed underscores to only show Ppl, Ppk, Ppu, Cp, Cpk, .. etc
                          • A new option – “Do not use unbiasing constant to estimate std dev for overall process capability indices” to compute overall Ppk (Ppl)
                          • Underlying charts (xbar.one) renamed to MR or I Chart based on SD or MR
                          • Handling of missing values
                          • Customizable number of decimals to show on the plot
                          • Standard Deviation label on the plot marked as “Overall StdDev” and “Within StdDev”

                          Process Capability Analysis for non-normal data

                          Six Sigma > Process Capability > Process Capability (Non-Normal)

                          A new dialog has been introduced to perform process capability analysis for non-normal data.

                          Multi-Vari graph

                          Six Sigma > Multi-Vari Chart

                          A new option has been added to adjust horizontal and vertical position offset to place/move the values for the data points on the plot.

                          Enhanced Shewhart Charts

                          Six Sigma > Shewhart Charts > …….

                          A new option has been added to all Shewhart Charts dialogs: the ability to add any number of spec/reference lines to the chart specified by the user.