Inspiration

Our project was inspired by the Colors of Van Gogh dataset and both team members' interest in Van Gogh's works. We wanted to understand how Van Gogh evolved throughout his career and discern relationships between elements of his art, as well as to see if we could make the archival and restoration process of paintings more accurate and efficient.

What it does

Our project conducts digital analysis on Vincent Van Gogh's artwork to identify interesting trends and patterns using various graphs. We also created an interpretable machine learning model that predicts the style of a Van Gogh painting.

How we built it

First, we came up with research questions that we wanted to answer and that our Colors of Van Gogh dataset could answer. We then decided the methods we would use to answer each question, reviewed the documentation for any new Python libraries we wanted to use, and split up the work by question. Lastly, we developed locally in Visual Studio Code before conducting analysis on our findings.

Challenges we ran into

One challenge we ran into was that we originally wanted to use dropdowns and interactive elements with the plotly library but found that we would need to filter our data in JavaScript, so we ended up using the bokeh library instead. Another challenge we encountered was that we had to adjust our machine learning challenge goal because we did not end up using the eli5 library to get feature importances.

Accomplishments that we're proud of

Some of the accomplishments we're most proud of include successfully learning how to calculate feature names and importances in sklearn, as well as successfully learning how to use bokeh to create stacked time series and graph bars in hex codes from our data.

What we learned

From this project, we learned a lot about local Python development (as we had to set up a local environment and learn about things like Anaconda to develop this project) as well as about various libraries. For instance, we learned more about the complicated graphs we can make using bokeh, as well as how to query an API in Python and process results using requests.

What's next for Van Gogh in the Age of Computers

In the future, we could expand this project by looking at most frequent colors as a percentage of total colors, rather than looking just at counts; by creating and training machine models similar to the one we created for Van Gogh but for other Post-Impressionist artists, to see if we can predict how other artists' paintings will be classified and also to see if the features and feature importances are similar to Van Gogh's; and finally, by querying the Met Museum API for data on other Post-Impressionist artists of the period and graphing their most frequent topics, to see how they compare to Van Gogh's.

Built With

Share this project:

Updates