InsightStack

InsightStack is a personal response to a recurring problem:
in development work, we talk about knowledge but rarely structure it.
Reports are scattered, templates are recreated from scratch, and institutional memory is fragile.
This repository is my attempt to change that.

InsightStack is a documentation-first, logic-driven knowledge repository built from years of work across research, evaluation, program design, and policy strategy. It brings together tools, templates, and methods that help answer one core question:

How do we build systems that learn, instead of start over every time?

Whats Inside

| Folder | Description | |--------|-------------| | observable_notebooks/ | Interactive data storytelling and dashboards using Observable | | open_data_editor/ | Drag-and-drop charting with RawGraphs, Datawrapper, and Open Data Editor | | taguette_coding/ | Qualitative coding projects for FGDs and transcripts | | visual_ethnography_descript/ | Narrative editing workflows using Descript | | kumu_maps/ | Systems maps and network diagrams for MEL planning | | stata_snippets/ | .do scripts for survey weights, DiD, and equity graphs | | excel_visuals/ | Gantt, waterfall, bullet charts for MEL reporting | | powerbi_reports/ | Example dashboard .pbix file and visualization | | spss_scripts/ | .sps syntax for tabulations, recodes, and exports | | flourish_charts/ | Storytelling dashboards using Flourish embeds | | excalidraw_frameworks/ | Hand-drawn MEL/system maps using Excalidraw | | survey_to_codebook/ | Convert XLSForms to Markdown codebooks automatically | | data_validation/ | Scripts for missingness, duplicates, value range checks | | label_variables/ | Label variables using dictionaries (R, Python, Stata) | | replication/ | Reproducible statistical analyses using synthetic data | | vensim/ | System dynamics models for development sectors | | docs/ | Architecture diagrams and visuals | | KM_tools/ | SOPs and templates for research documentation |

What Its For

This repo is for anyone working in: - Public health, climate, education, or WEE programs in South Asia - MEL teams building documentation systems that go beyond Excel trackers - Researchers and analysts writing reports that need to live longer than a funding cycle - Program leads tired of reinventing logic models and tagging templates

How to Use

• Clone or download the repository
• Explore individual folders everything is open, documented, and editable
• Plug in your own data into calculators or adapt them to your setting
• Use the SOPs in KM_tools/ to create your own tagging systems, documentation trackers, and research logs
• Contribute back via pull requests or suggestions

Data Flow

A typical workflow in InsightStack looks like this:

1. Raw data is collected from XLSForms or CSV inputs
   
2. Validation scripts detect missing or invalid values
   
3. Variables are labeled for readability using lookup dictionaries
   
4. Data is analyzed using standard R/Python scripts
   
5. Results and tools are documented or packaged for replication

See data_flow.md
View the InsightStack Flow Diagram

Tools Used

InsightStack includes real, working scripts using:

• Python: pandas, statsmodels, pyreadstat, Jupyter
• R: haven, labelled, dplyr
• Stata & SPSS: Labeling, validation, and transformation
• Other: XLSForms, Markdown, Miro, Excalidraw, LaTeX

See tools_used.md for a full list

Citation

If you use InsightStack in your work, please cite it as:

Varna Sri Raman. (2025). InsightStack: Knowledge Documentation for Development Research [Data & Tools]. Zenodo. https://doi.org/10.5281/zenodo.15245182

Repository Status & Metadata

<!-- Paste export-to-bibtex and ORCID badges here -->

Related Repositories

• EquityStack: Modular Python workflows for public health, WEE, MEL, and education data analysis
  
• FieldStack: Applied R templates and calculators for social sector research and dashboard-ready outputs
  

Contact

Want to contribute or adapt it for your own org?
Drop a message via GitHub issues or email me: varna.sr@gmail.com

License

This repository is licensed under the MIT License. Use freely, adapt creatively, credit generously.

Changelog

You can view all updates and commit-based changes in the CHANGELOG.md.

This file is automatically updated via GitHub Actions.

2025 Varna Sri Raman. Built for open knowledge, shared learning, and programmatic memory.