Our first day of Dashboard Week required us to connect to the API for Cooper Hewitt, Smithsonian Design Museum to find some data and make a dashboard to show a story we found.
I began by reading the API documentation and choosing to look at the exhibition and object data to find any trends in the number of active exhibitions and number of objects being shown over time.
In Alteryx
I connected to the exhibitions.getList API to obtain a list of each Cooper Hewitt exhibition and all relevant associated data.
I then fed each of the exhibition IDs into the exhibitions.getObjects API to obtain a list of all objects in each exhibition and all relevant associated data using a batch macro.
Finally, I used the departments.getList API to obtain a list of each Cooper Hewitt department to replace the department ID with the department name in the objects data.
After obtaining all relevant data, I did some data preparation including calculating duration of each exhibition and fixing some data quality issues. I also made a dates table from the minimum exhibition start date and the maximum exhibition end date for visualisation purposes.
In Tableau
Now that I had all my data, I focused on representing the number of active exhibitions and objects on display over time. It was apparent that there was a sharp increase in active exhibitions from 2014 onwards so I planned to orient my dashboard around comparing until 2014 and from 2014. I tried investigating a few possibly interesting relationships and became most interested in a scatterplot which determined whether there was any relationship between the duration of an exhibition and the number of objects it featured. Additionally, I thought it was interesting to break the objects up by type and see if there were any trends. Finally, there were some more recent exhibitions that displayed as having very few objects (in some cases 0). Unfortunately, it looks like some of these exhibitions did have a number of objects on display but they may have not been digitised yet or the exhibition records may not be updated, so I added a parameter to eliminate data with too few a number of objects. I hope you enjoy the result!
