Data created in historical periods is extremely limited, but data scientists can create new data about the past in order to draw new insights though data analysis and machine learning. In this project, I use a dataset compiled by a group of historical experts about societal normas across time in different areas to answer three questions, outlined below under Summary of Analysis Results. Final analysis can be found in the associated blog post.
- Numpy
- Pandas
- MatPlotLib
- Seaborn
- SciKit-Learn
- README.md README file informing users of the project summary, dependencies, files in the repository, a summary of results, and acknowledgements.
- empires_and_eda.ipynb Jupyter notebook with the code for my analysis.
- data/axial_age.csv CSV dataset of the presence or absence 12 sociopolitical atributes over 10 geographic areas for each 100 years, ranging from the years 5300 BCE to 1800 CE.
This research employed data from the Seshat Databank (seshatdatabank.info) under Creative Commons Attribution Non-Commercial (CC By-NC SA) licensing. Turchin P. et al. 2015. “Seshat: The Global History Databank.” Cliodynamics 6(1): 77–107 Mullins, D., Hoyer, D. et al. 2018. “A Systematic Assessment of ‘Axial Age’ Proposals Using Global Comparative Historical Evidence.” American Sociological Review 83(3): 596–626
In this notebook I answer the three following questions throught the corresponding methods outlined below:
- What sociopolitical norms did different societies from across the Mediterranean and Asia develop over time in comparison with each other? Comparative barplots
- Which of these norms were most and least likely to be observed in the same place and time? Correlation matrix
- What are some of the distinct trajectories of related norms along which some groupings of societies but not others evolved? Principal Component Analysis
Special thanks to Seshat for making several datasets from their global history databsnk publically available, and specila thanks to Udacity, whose rubric inspired this project and without whom my PCA would have taken much longer.