|13 Mar 2023||13:00 - 17:00||IT Training Room, University Library, West Road, Cambridge|
Convenor: Estara Arrant (CDH Methods Fellow)
This methods workshop will teach students three powerful machine learning algorithms appropriate for Humanities research projects. These algorithms are designed to help you identify and explore meaningful patterns and correlations in your research material and are appropriate for descriptive, qualitative data sets of almost any size. These algorithms are applicable to virtually any Humanities field or research question.
- Multiple Correspondence Analysis: automatically identifies correlations and differences between specific data elements. This helps one to understand how different features (‘variables’ or ‘characteristics’) of one’s data are related to each other, and how strong their relationships are. This can be useful in almost any research project. For example, in a sociological dataset, this analysis could help clarify relationships between specific demographic characteristics (race, gender, political affiliation) and socioeconomic status (working class, education level, income bracket).
- K-modes clustering and hierarchical clustering: finding groups of similarity and relationship within the entirety of your data. Clustering helps one to identify which variables/characteristics group together, and which do not, and the degree of difference between groups. For example, such clustering could allow an art historian to see how paintings from one decade are characterised by style and artist, as contrasted to paintings from another decade (thus tracking shifts in artistic trends over time).
This workshop will specifically cover the following:
- Determining when your research could benefit from machine learning analysis.
- Designing a good methodology and running the analysis.
- Interpreting the results and determining if they are meaningful.
- Producing a useful visualisation (graphic) of the results.
- Communicating the findings to other scholars in the Humanities in an accessible way.
Students will actively implement a small research project using a practice dataset and are encouraged to try out the methods in their current research. They will learn the basics of running the analysis in R’s powerful programming language.
Target audience: This course is appropriate for Humanities scholars who are beginners to programming and data science and those with some experience. CDH Methods Workshops are open to staff and graduate students who want to learn and apply digital methods and use digital tools in their research. Participants are requested to complete this simple information questionnaire before the event.