This repository provides my solution for the 2nd Assignment for the course of Practical Data Science for the MSc in Data Science at Athens University of Economics and Business.

Replicate some of the findings of a well-received book, Why Civil Resistance Works by Erical Chenoweth and Maria J. Stephan. The two authors examined whether nonviolent campaigns are more effective than violent campaigns. They found that yes; from a data set containing details of campaigns from 1900 onwards, they found that in general peaceful campaigns are more effective in achieving their aims than violent ones.

To answer the following questions we will use the data that the authors helpfully provide at https://www.ericachenoweth.com/wp-content/uploads/2012/01/Data-and-Replication-Files-3.zip. The data is in Stata format, which pandas can read without a problem.

Creating a plot showing the frequency of nonviolent and violent campaigns; the frequency is counted based on their end years, at decades. The violent and nonviolent campaign counts are stacked on each other.

Creating a plot showing, for each decade from 1940 onwards, the number of nonviolent campaigns and the percentage of success.

Creating a plot showing the success rate for violent and nonviolent campaigns by decade.

Creating a table with the twenty-five largest resistance campaigns, for all years, based on their number of members at highest point (peak membership).

Examining the effect of participation on the probability of campaign success.

From the results obtained, we find the percentage of the population that is the tipping point for success in a campaign.

Examining whether nonviolent resistance remains significant even if we control for the regime type and also whether nonviolent resistance remains significant even if we control for geographical location of the compaign.