Concept and Dataset Procurement
I am interested in coming up with new ways large data sets can be brought to life. As an archivist and a student of library science and art history, I am passionate about the history of photography. I want to investigate the circumstances that have led to the establishment of photography as an “accepted high art form” and that conception’s subsequent influence on the sales of photography at auction.
In order to analyze photography collecting trends, I have used python to scrape the results of six Dreweatts & Bloomsbury auctions that exclusively sold photographs. The availability of this information online made this project possible. Otherwise I would have had to of cross referenced the results of the auctions with the catalogues to compile this dataset.
Through a series of python scripts I was able to gather information about 1,415 photographs that were sold between 2010 and 2014. You can see the entire programming element of the project on github. The dataset for this information visualization project is comprised of: the name of the artist, the year in which the photograph was created, and the price for which the photograph sold or whether it did not sell. I then adjusted the prices for inflation, and since Dreweatts’ prices are in Pounds, I converted the amount to Dollars and adjusted for date specific conversion rates. In order to work at a more manageable level of trend analysis I used the date the photograph was created to generate the decade in which the photograph was created.
Visualizations
I wanted to visualize this data set to answer a few specific questions. What decade was the most prevalent at auction? Which decade generated the most revenue? Did prices at which the photographs sold increase from 2010 to 2014? Did famous photographers works’ sell at higher prices than less prominent photographers? The results of this study are not able to be extrapolated to industry wide trends since the dataset is by no means large enough to be indicative. However, this project is one step towards understanding trends in the photography market.
Using Tableau software, I imported my dataset and created visualizations to aid in understanding photography auction trends. In order to judge the results of the auctions, I first needed to look at the amount of photographs that were available by decade in each auction year. The following graph shows this distribution.
From this graph we can see there were relatively few photographs created during the 1800s up for auction. This makes sense because photography was only invented around 1820. Therefore the medium was not as prevalent, producing few images at a great cost. One of my assumptions was that the photographs up for auction during this time would sell for a large sum of money since they are scarce in comparison to the proliferation of images in the past century. Later on we will see how wrong I was. Turning back to the graph, we can see the sharp increase in photographs available from the 1950s onwards. This was the result of many technical developments in photography: faster film speeds, the advent of color and flash photography, more affordable and user-friendly cameras, etc. You can view the interactive graph, where you can see the exact number of photographs available by auction year, by following this link.
The next visualization shows us the total amount of revenue generated by the sales of photographs broken down into the decade in which the photographs were created.