Graffiti is everywhere - from the low-lying corners of a university bathroom to the soaring heights of a highway overpass. No matter where it is found, the interplay of colors, techniques, style, and wordplay act as a vortex - guiding eyes into a world of abstract creation and interpretation. Based on the sheer magnitude of graffiti present throughout Seattle, it is evident that this community has survived and continues to be quite active. From this, one may ask: What are key patterns exhibited by graffiti artists that are displayed through data analysis? Are there locations writers choose over other spots? Is the type of graffiti hindered by a quality-vs-quantity relationship? Many of these are explored to better understand writers' behavior and intent. By using landscape archaeology and data-science techniques, the observed patterns of three writers have been analyzed and further studied to understand canvas location, graffiti type, color use, visual composition, and other related components. By using data collected from the ARCHY 235: Explore Graffiti course, I have been able to culminate and process individual entries through the computing platform Jupyter Notebooks. Through Python coding, I have been able to derive maps, various descriptive statistics, and charts to help gain insight into each artist's work. Furthermore, by focusing on three specific and well-documented writers, commonalities and distinctive traits have been uncovered. An example of these findings includes the use of high-contrast black and white markers. These are used because of their possible reflective capabilities. Additionally, stickers are popular because artists can mass produce their tag and dramatically reduce the time it takes to leave their mark while significantly decreasing the chance of being caught. Thus, as a result, the study has shed light on this sociological phenomenon by focusing on examining and investigating individual practices exhibited by these artists through their graffiti.

Ceramic petrography involves the examination of the mineralogical and microstructural composition of ceramic artifacts known as sherds. Traditionally, this examination to generate quantitative data from the ceramic material is often done by the hand, particularly the examination of ceramic inclusions in both their size, amounts, shapes, and density. This process is time-consuming, error-prone, and leads to limited datasets which are difficult to compare. By leveraging Python imaging analysis capabilities, a standardized and rapid method to generate repeatable datasets can be achieved on a greater scale. From work previously performed by my mentor, Alec Iacobucci, Alec created a robust set of internally consistent functions to collect multivariate data from ceramic sherd scans including inclusion size and count. This was done through Scikit-Image, an open-source image processing library for Python. However, Scikit-Image has its limitations, specifically in its computational efficiency as processing times are long and extensive. My goal for this project is to move the current material we have in Scikit-Image and translate it into OpenCV which is another library for image processing, but OpenCV is better optimized and specializes in computer vision. I will match similar functions and capabilities between Scikit-Image and OpenCV, perform data analysis on the ceramic inclusions to see the degree in which OpenCV outperforms Scikit-Image, and expand on previous determined variables such as average color of inclusions, orientation, and shape. The broader implications of this research is to reduce processing time of inclusion counting in the field which not only allows for larger and standardized datasets that can be compared and analyzed but to automate the ceramic petrographic process beginning at the dig site all the way to data analysis. This will be through the creation of an on-site scanner that implements our algorithms that is both portable and efficient where researchers can carry with them. 

Graffiti surrounds us everywhere we go. It can be found throughout Seattle, ranging from small tags or stickers that are just about anywhere to large pieces on walls and the sides of buildings. Even though it is so widespread, the locational aspects of graffiti remain unexplored. The data used in this project was collected by students using a mobile phone app as part of the fall 2021 course Archaeology 235: Exploring Graffiti: Combining Landscape Archaeology and Data Science. This study attempts to determine whether different types of locations emerge when attending to various locational factors (e.g., viewing potential and accessibility). To this end, I use various multivariate techniques (e.g., clustering, multiple correspondence analysis) utilizing various Python packages (Pandas, Scikit-Learn). These techniques reveal natural clusters within the data by grouping data points that are more similar to each other than they are to data points in other groups. Each technique does this differently. The goal of this research is to explore how locational factors of graffiti interact with each other and to investigate which clustering techniques are best at uncovering these interactions. The findings of this research show how different aspects of graffiti are associated with one another, allowing me to understand how various factors influence graffiti production. With this understanding, we can study Seattle’s graffiti scene on a larger scale.