This repo contains code in which we analyse the causal effects of various features in Kickstarter campaigns. Here is an example of a featured campaign.
Please see the final report for the full writeup of our findings.
The script in
python/kickstarter_main.py
Contains the methods to collect, clean, and engineer features from the raw data obtained by a web crawler here
To merge the raw csv files together from webrobots, run
# Example
python3 python/kickstarter_main.py --mode merge
To collect additional information form the campaign urls directly, run
# Example
python3 python/kickstarter_main.py --mode collect --batch 5 --wait 250 --max_request 200
To preprocess and clean features, run
# Example
python3 python/kickstarter_main.py --mode features
Rmd file for plotting covariates and looking at various interesting tables summarizing the data can be found in
R/Kickstarter_Causal_Analysis.Rmd
Rmd file for preparing the data for casual analysis on three chosen treatment variables and performing hypothesis testing under Fisher and Neyman frameworks.
R/Hypothesis_Testing.Rmd