Check this out! (Code here.)

Government budgets are opaque. They utterly, absolutely, completely dwarf smaller social sector organizations trying to make anything better. They are also poorly understood.

This is an attempt to demystify them - mostly for myself, maybe for others. This is also a WIP, and will (may?) get better over time. Understanding these datasets better helps us understand entrenched interest groups that have captured a disproportionate share of these sums (a typical pattern in democracies).

Some quick notes on the process:

  1. Indian govt. is organized as ministries, which are internally organized as departments. Some departments are directly managed by the PMO (I think?) and do not have a parent ministry as such. Each ministry creates multiple demands for grant, one for each department.

  2. I parsed a giant Excel sheet containing 100+ of these demands for grants, and organized them into a tree-like structure. This process was painful, and the python code that does this is a reflection of that. There are a couple of places where things get rounded off due to heuristics breaking down, but they should not cause massive issues.

  3. The data visualization in in a decent JS stack (NextJS, React, Typescript, Plotly.js …). Feel free to contribute!