MAX – Museum Analytics
Authors: Stefanie Schneider
Working with Shiny more than 1 year
Abstract: Museum Analytics, MAX, is intended to enable lecturers to import freely selectable museum databases and make them available to students for analysis. The aim is to provide an interface that facilitates the introduction to programming with R through an intuitive visual feedback system.
Full Description: In recent years, large museum databases have been created in the international museum sector that are awaiting meaningful use. They offer a hitherto unknown opportunity for empirical investigation of the history of collections, which can be expected to yield far-reaching results, especially in a comparative perspective. Museum Analytics, MAX, is intended to enable lecturers to import freely selectable museum databases and make them available to students for analysis. The aim is to provide an interface that facilitates the introduction to programming with R through an intuitive visual feedback system. In so doing, students acquire knowledge in the statistical analysis of relatively large amounts of data; an ability that is becoming increasingly important in a museum work context, but which is virtually ignored by classical art history.
First, either load one of the predefined data sets or import your own (Screenshot 1). Your currently selected data set is displayed on the left, either as a table or a plot (2). You can now preprocess and visualize this data, e.g., standardize dates or draw a boxplot. If you do not want to separate preprocessing and visualization, you can also do both in one window (3). On the right you can define tasks to be performed on your data set. First, either add a new task or import a file with tasks from a previous session. Each task can be further specified, e.g., you can temporarily disable it; or view the associated documentation (4). Finished? Let’s run the selected tasks to see if they can be completed successfully (5). If not, the respective task is marked yellow (a warning has occurred) or red (an error has occurred). The processed, cleansed, and visualized data can be exported as a
.zip file (6).
Initially, 17 institutions with almost 500,000 objects were selected as museum collections to be integrated into the tool: e.g., the National Gallery of Art and the Städel Museum in Frankfurt. These were extracted by means of web scraping. Since all selected collections are annotated with the iconographic classification system Iconclass, they can be meaningfully examined in humanities seminar contexts.
- A graphical user interface that enables fast progress without excessive training. Each task can be further specified, e.g., can be temporarily disabled. The order of the tasks can be changed with drag & drop.
- An import module to “pull” existing data, e.g., from the Rijksmuseum in Amsterdam or Österreichische Galerie Belvedere, into the tool as easily as possible. Own data sets can be fed in just as easily. Currently supported are
- An export module to extract the processed, cleansed, and visualized data as a
.zipfile with R-compatible
.rdsfiles. Reproducible R code can also be generated based on the defined tasks.
- Interactive tables with DataTables that allow to select rows by setting local and global filters. A flexbox layout is used to display the relevant section of the table always next to the tasks to be performed.
- Dynamic and interactive graphics with Plotly that show more details on mouseover, e.g., the title or artist of an artwork. They enrich the statistical analysis by displaying complex relationships in an attractive way. Plot subregions can be selected and zoomed in.
If you want to extend the functionality of MAX, you can add an R package function to the
.yaml file in the folder
data that corresponds to the respective section, i.e., currently either
visualize-history.yaml. The function to be added must be Pipe-friendly or ggplot-compatible.
About the Project
MAX was funded from 1 March to 30 November 2018 within the program Lehre@LMU to strengthen research orientation in teaching. It is a project of the IT-Gruppe Geisteswissenschaften, the Institute of Statistics and the Institute of Art History at Ludwig-Maximilians-Universität München. Our team consists of Severin Burg, B.A., Prof. Dr. Hubertus Kohle, Prof. Dr. Helmut Küchenhoff and Stefanie Schneider, M.Sc.
The web application is written using R and the Shiny web framework. It is open source and licensed under GNU General Public License v3.0. This version is a complete re-implementation that makes use of Shiny modules and custom HTML templates. For the previous version, please see: https://dhvlab.gwi.uni-muenchen.de/max/.
Keywords: Exploratory Data Analysis, Data Visualization
Shiny app: https://sschneider.shinyapps.io/museum-analytics/
RStudio Cloud: https://rstudio.cloud/project/1077562