This repository contains all code to generate the results for the paper 'Incorporating Machine Learning into Sociological Model-building'.
The repository contains code in five separate folders. Each folder contains reproduction material for an example used in the paper. The first folder base_simuls contains a script to generate a couple of non-linear functional forms. Each of the other four folders has the same script structure. First a 00_gen_data.R script used to wrangle the data. 01_superlearner.R estimates typical functoinal forms together with a SuperLearner containing various flexible models. 02_simul.R evaluates the out-of-sample fit metric of the best performing flexible model from the SuperLearner and the hypothesized model at different feature sets. 03_shap.py calculates Shapley values for the flexible model. 04_plot.R plots the results.
More detailed information for the housing, mincerian and ideology examples is enclosed in a folder-specific README.md.
├── base_simuls
│ ├── 00_gen.R
├── toy_example
│ ├── 00_gen_data.R
│ ├── 01_superlearner.R
│ ├── 02_simul.R
│ ├── 03_shap.py
│ ├── 04_plot.R
├── mincerian
│ ├── README.md
│ ├── 00_gen_data.R
│ ├── 01_superlearner.R
│ ├── 02_simul.R
│ ├── 03_shap.py
│ ├── 04_plot.R
├── housing
│ ├── README.md
│ ├── 00_gen_data.R
│ ├── 01_superlearner.R
│ ├── 02_simul.R
│ ├── 03_shap.py
│ ├── 04_plot.R
├── ideology
│ ├── README.md
│ ├── 00_gen_data.R
│ ├── 01_superlearner.R
│ ├── 02_simul.R
│ ├── 03_shap.py
│ ├── 04_plot.R