Markers of Inequality (MOI): Intersectional Needs Assessment

This repo contains code to study intersectionality in needs assessment for development planning. It accompanies the paper "Shedding light on vulnerability: Intersectional energy planning for development".

This method was developed using a case study dataset collected in Siaya County, Kenya. This dataset covers, amongst other things, data on which items people prioritize most in their daily lives and the reasons why (i.e., their values). More information on the case study dataset is available here.

How to run the analysis

To run the analysis, first ensure that Siaya_UPV_survey.csv and Siaya_UPV_Utterances.csv are in the \data folder. These can be retrieved from the dataset link if needed. Then, run the data pre-processing by issuing the following command from the MOI directory:

python preprocess_data.py

This will generate two new files, Siaya_UPV_survey_Prepared.csv and Siaya_UPV_Utterances_Prepared.csv. These files contain one-hot columns for each demographic marker to evaluate. Now, you can run any of the analysis scipts from the MOI directory using a command structured as:

python <script_name_here>.py

About the analysis scripts

Each script does a different type of analysis. There are generally two types:

1. Map the intersectionality of the sample (i.e., intersectionality.py)
2. Map the prioritization of items across demographics (i.e., item_matrix_<x>.py)

The <x> contains flags about the type of analysis in the script. These are:

• intersections vs singlemarker: The scripts which include "intersections" analyse each intersection of two demographics, while those which include "singlemarker" analyse each individual demographic.
• climate vs general: The scripts which include "climate" address the climate-event-specific needs, while the scripts which contain "general" address needs in daily life.

These scripts output both heatmap matrices and spreadsheets summarizing intersectional item prioritisation.

How to customize the analysis

While these scripts are tailored to the sample dataset, they can be modified for any dataset including demographics which can be one-hot encoded and needs-related data. This can be done by editing the pre-processing file to create one-hot columns for demographics of interest, and replacing the item/value lists/references with the relevant needs-assessment datapoint in your own dataset throughout the code.

Citation

If you make use of this codebase, please use the following citation:

Leonard, A., Nguti, K., Lanza, M. F., & Hirmer, S. (2025). Shedding light on vulnerability: Intersectional energy planning for development. Renewable and Sustainable Energy Reviews, 211, 115199. https://doi.org/10.1016/j.rser.2024.115199.

@article{leonard2025shedding,
  title={Shedding light on vulnerability: Intersectional energy planning for development},
  author={Leonard, Alycia and Nguti, Kuthea and Lanza, Micaela Flores and Hirmer, Stephanie},
  journal={Renewable and Sustainable Energy Reviews},
  volume={211},
  pages={115199},
  year={2025},
  publisher={Elsevier},
  doi={10.1016/j.rser.2024.115199}
}