georank 0.1.0

Generic weighted-criteria ranking engine — score and rank places (or anything) by normalized, weighted factors. Zero dependencies.

georank

Generic weighted-criteria ranking engine. Score and rank places — or anything — by normalized, weighted factors. Zero dependencies, available for both JavaScript and Python.

This is the open-source core of the ranking method behind GeoRank.place, a map-based tool for comparing cities and regions on the factors that matter for relocation (sunshine, cost, climate, and more).

Method

Each criterion is min-max normalized to 0–100 across the set you pass in, then weighted and summed into a single 0–100 score:

1. For every criterion, the lowest raw value maps to 0 and the highest to 100.
2. Set invert: true for criteria where lower is better (cost, tax, rainfall).
3. Missing values map to a neutral 50 rather than penalizing an item to zero.
4. Weights need not sum to 1 — they are normalized internally.

Install

npm install georank      # JavaScript / TypeScript
pip install georank      # Python

Usage

JavaScript / TypeScript

import { rank } from "georank";

const cities = [
  { name: "Lisbon",   sunshine: 2799, cost: 100 },
  { name: "Berlin",   sunshine: 1626, cost: 85  },
  { name: "Valencia", sunshine: 2696, cost: 70  },
];

const ranked = rank(cities, [
  { key: "sunshine", weight: 0.6 },
  { key: "cost",     weight: 0.4, invert: true }, // cheaper scores higher
]);

ranked.forEach((r) => console.log(r.item.name, r.score, r.signals));

Python

from georank import rank

cities = [
    {"name": "Lisbon",   "sunshine": 2799, "cost": 100},
    {"name": "Berlin",   "sunshine": 1626, "cost": 85},
    {"name": "Valencia", "sunshine": 2696, "cost": 70},
]

ranked = rank(cities, [
    {"key": "sunshine", "weight": 0.6},
    {"key": "cost",     "weight": 0.4, "invert": True},
])

for r in ranked:
    print(r["item"]["name"], r["score"], r["signals"])

API

rank(items, criteria)

Returns a new array sorted by score (descending). Each result is { item, score, signals }, where signals holds the per-criterion normalized 0–100 value.

criteria is a list of { key, weight, invert? }.

normalize(values, { invert })

Min-max normalize an array of numbers to 0–100. Missing / non-finite values become 50; if every value is equal, every output is 50.

Live tool

See the method applied to real climate and cost data at georank.place — including the interactive city rankings and side-by-side compare.

License

MIT © GeoRank.place