Chapter 1 Introduction

1.1 Data

The data I plan to use is dataset from the Metropolitan Museum of Art’s collection. To use this, I plan to access the API, which can be found at this link: https://metmuseum.github.io/. I plan to use artworks that are in the Met’s collection and their data to predict whether the Met deems them important.

This can be done by analyzing how I can predict the “isHighlight” variable. The isHighlight variable when “true” indicates a popular and important artwork in the collection”, according to the Met’s API documentation. An example of one such work is Van Gogh’s “Wheat Field with Cypresses”.

1.2 The Modeling Goal

The modeling goal of this project is to predict the isHighlight variable and use it to reveal biases in the Met’s decision making of what is an important or non-important work. Using features such as the department of the work (eg. Egyptian Art), the culture of the art, the artist’s demographics, the artist nationality, medium, dimensions, country created in, region, and general classification.

1.3 Models I Intend To Use

I plan to use the following models
* a simple OneRule model
* decision tree
* random forest
* support vector machine
* artificial neural network

I plan to create all of these models with the intention of showing the pros and cons of interpretability. In addition, I wish to see if the different models achieve accuracy using different predictors or if they rely on the same features.

I expect that the first two will be the most interpretable, while the last three will have the highest accuracy.