Part 2. The FLOAT Method
2.5 Analyze
Garnering Findings from Data
Data analysis is becoming an increasingly important part of digital humanities work, and three particular areas of analysis—exploratory data analysis (EDA), data visualizations, and statistical computing—are central to the digital humanities. The Chapter 2.2 Formulate covered EDA and data visualization. Therefore, this chapter will continue what was discussed, going into statistical computing.
Analyze
Data analysis is a broad term that encompasses any method of using data to come to a finding. This can be as simple as counting things. Visualizations of data, summary tables, and descriptive statistics are all used to show findings of varying amounts of complexity and importance.
For example, the article “Reporting and Availability of COVID-19 Demographic Data by US Health Departments (April to October 2020): Observational Study” is an analysis of data being provided by health departments about COVID-19 during the pandemic. While some statistical tests were run, the majority of the findings are displayed on a three-part map, seen below, showing what types of information are being shared with the public over time.
Map of COVID-19 Data Availability, October 2020
Pivot Tables
Pivot tables are a method of summarizing datasets. It is based on turning categories into averages, sums, counts, and other statistics, making it useful to recognize patterns. For example, looking at the file “Beyonce – Awards.csv” in the “Storytelling with Data – The Beyonce Edition” dataset, we see the following variables: Song Title/Album Title, Year, Award Institution, Award Category, Award Type, International Award, and Result (see image below). These are listed one nomination at a time, totaling 651 rows. It is difficult to come to any conclusions scanning these data.
Beyonce Award Data from “Storytelling with Data – The Beyonce Edition”
Beyonce Award Pivot Table
In order to see totals and averages of these data, the categories can be dragged and dropped onto a pivot table. Relationships can even be discovered. When placing some of these data into pivot tables, we can draw conclusions.
Looking at the first pivot table of awards received by year, we observe she has been fairly consistent in being awarded, receiving an average of 12 awards per year since 2001, with her highest awarded years as 2015 and 2017 with 25 and 29 awards received.
In the second table, when looking at all her nominations, she seems to win 2 out of 5 (or 262 won out of the 650 grand total).
In the third table, we can break things up to see if anything stands out when looking at categories according to race and gender. When looking at categories based upon typical nominees and awardees: For Gender & Race awards (such as Best Female R&B) and Race awards (such as Best Urban Song), she wins about half the awards she’s nominated for, but looking at Gender awards (such as Female Artist of the Year), she wins a lot fewer (19) than she was nomina