A2-Mackenzie Leake

Introduction and Initial Questions

When I visit art museums, I am always curious how the museums select pieces to acquire and display. An extensive body of literature surrounding curatorial practice explores these complex decisions. For this project I decided to perform my own investigation into the question: How do art museum collections grow and shift over time? This is a very broad question, so to help narrow my focus I came up with a list of some of the ways that a collection could shift over time:
- Artists represented
- Titles of the pieces
- Mediums of the pieces
- Sizes of the pieces
- Number of pieces acquired each year

Based upon this list I formulated several questions pertaining to how the artwork in a collection might shift over time. Here are three of the questions I came up with:
1. Do museums acquire more pieces during good economic times?
2. How long do museums wait after the completion of a piece to acquire it?
3. What sizes and shapes of artwork are most commonly acquired?

The Data

At this point I wanted to leave my options open for exploring any of these questions. I started searching for museum data sets. I was able to find individual datasets for different museums but no dataset across multiple museums. The Tate Collection provided a dataset in a CSV file with nearly 70,000 entries for artworks acquired between 1823 and 2013 (https://github.com/tategallery/collection). I originally looked through the data in Excel, and then I performed all of the rest of my exploration in Tableau. I decided to use this dataset to explore: How have the Tate's acquisitions shifted over time?

This dataset provided information for each acquisition, including its:
- Title
- Medium
- Height
- Width
- Acquisition year

Exploratory Process

I then revisited my three questions from above to see which of these I could answer with the available data. I realized I could not answer question #1 with this dataset alone because I had no economic data available. The dataset did, however, appear to provide opportunities to explore questions 2 and 3.

I started to answer a form of question #2: How long does the Tate wait after the completion of a piece to acquire it? I went into Excel and added a new variable: time difference, which is the difference between the completion date of the piece and its acquisition by the Tate. I imported the spreadsheet with this additional information into Tableau. I began by plotting the number of pieces with each time difference:

Here was the full visualization:

I realized that there were too many separate categories so I then divided the time differences into bins:

This visualization allowed me to see that most of the works were acquired by the Tate within 60 years of their original creation. I could speculate that this is at least in part due to the Tate's interest in recent years in collecting modern and contemporary art.

I next decided to explore the data further by pursuing a form of question #3: What sizes and shapes of artwork are popular in the Tate Collection? In order to answer this question, I realized that I would need the height and width values for each piece. After filtering out null values for the heights and widths, I plotted height on one axis and width on the other axis. This data would not tell me much about the actual shape of the object, i.e., whether it was round or rectangular, but it would let me see the relationship between the heights and the widths of the pieces. I constructed the following visualization to explore this relationship:

While this visualization does allow us to learn more about how the heights and widths of artworks in the collection are related, I decided that a far more interesting question would be: How do the sizes of acquisitions change over time? I hypothesized that over time the Tate would start collecting smaller pieces as it might be facing tighter space demands. To answer this question I then added the dimension of acquisition year and filtered out null values in the dataset. I decided to examine median heights and widths for pieces acquired each year. I chose to use the median instead of the mean because many years had few acquisitions and there is no indication that the data is normally distributed. I constructed the following plot to show how the heights and widths of pieces acquired by the Tate have changed over time:

I decided that there was little benefit in showing the heights and widths separately. In Excel I updated the spreadsheet with a new variable: "dimensions," which is the product of the height and the width. This allowed for a single value to represent size. I then constructed the following visualization in Tableau:

Before preparing the final visualization, I changed the size units from mm^2 to m^2 in order to make the scale more understandable to the viewer. I also updated the caption for the final version.

Final Visualization

This visualization has been designed in Tableau to help answer the question: How have the sizes of artworks acquired by the Tate Collection varied over time? To explore this question, the median size, i.e., the height multiplied by the width of each acquired piece, is shown in relation to the year that it was acquired. The red line on the graph traces how the median size of acquisitions has varied from 1823 to 2013. This line allows us to see that the size of acquisitions peaked during the 1860's and has generally decreased since then. This visualization intentionally does not provide information about the number of pieces acquired each year. Rather, it shows the median acquisition size for each year to convey the size of a typical acquisition for a given year.

I originally hypothesized that the size of the pieces would decrease over time. My intuition was that over time galleries and storage facilities might become too crowded, and therefore curators would seek smaller pieces. It does appear that the size of acquired works has tended to decrease over time, but this visualization provides no insight into the cause of this decline. Is lack of storage to blame? Have trends in the art world led to artists creating smaller works? Have different curators throughout the years expressed varying size preferences? Although this visualization cannot answer these questions, it does help us see the relationship between acquisition year and the size of the artwork. This opens the door for further research into the the dynamics of this relationship.