Storytelling for exploratory data analysis (of Bitcoin)

This paper presents a method of storytelling as a way for an analyst to improve understanding of a dataset.  In the paper we explain our experience developing a story for an exploratory data analysis on the bitcoin blockchain.

Exploratory Analysis of Bitcoin Data

Abstract—The process of analyzing a set of data of which one is not previously familiar is an open-ended challenge that presents many potential options to the explorer. We present a method that emphasizes the explorer’s self-education in the domain via original story creation. The creation of a metaphorical story whose elements mimic real-life scenarios relevant to the data can helps in two ways: 1) as a method to aid the explorer’s understanding of different aspects of their dataset, 2) to improve the communication and following discussions of the data analysis.

To help develop this process, we conducted an exploratory data analysis on a subset of bitcoin activity. We followed a basic high-level process for the analysis: 1. Getting to know the domain, 2. Creating an overview of the data, and 3. Diving deep on select questions formed during the first two steps. We hope that revisiting elements of the story throughout the investigation will reaffirm the strength of our metaphor, and to aid in the discussion and creation of new and interesting avenues to explore.