How each factor is placed: rank, distance from typical, lowest to highest
Cafés are counted in units, population in people. Before a score can add them, every factor has to be placed on one scale. 'Where it ranks', 'How far from typical' and 'Lowest to highest' are three honest ways to do that, and they give different maps. Here is each one on London.
A score blends factors measured in different units: 23 cafés, 1 891 residents, 6 transit stops. Before any of that can be combined, each factor has to be turned into a position between 0 and 1 - where does this area sit, for this factor, among all the others? Only then can positions be weighted and blended.
There is more than one honest way to do that, and they give different maps. How each factor is placed is the control that chooses. Its three options - “Where it ranks”, “How far from typical” and “Lowest to highest” - are what the field calls percentile rank, z-score and min-max normalisation. Analysts argue about these; here they are one dropdown, in plain words, and this post is about what each one does to a London map.
How each factor is placedthe field's term: Normalisation (rank · z-score · min-max)
Cafés, residents and transit stops, each placed by 'How far from typical' and then combined. Neighbourhoods genuinely above typical on all three glow; ordinary ones sit in the middle.
The three ways, side by side
Where it ranks
Sort every area from lowest to highest and place each by its position in the queue. The median sits at 0.5, the top at 1. One freak area cannot stretch the scale. The size of a gap is lost.
How far from typical
Measure how many spread-widths (standard deviations) an area sits from the typical value. A big gap reads as a big gap. Three spread-widths either side become 0 and 1; anything beyond is capped.
Lowest to highest
Place each area on the straight line from the smallest value to the largest. Gaps keep their size exactly. One extreme area stretches the whole scale and squashes everyone else.
Five areas, cafés counted: 1, 2, 3, 4 and 40
Where it rankseven steps, whatever the values
0 · 0.25 · 0.5 · 0.75 · 1
How far from typicaltypical = 10, spread ≈ 15, mapped from ±3 spreads
0.35 · 0.36 · 0.36 · 0.37 · 0.86
Lowest to highest
0 · 0.03 · 0.05 · 0.08 · 1
The 40-café area is the same in all three. What changes is everyone else: ranked, they are spread evenly; by distance from typical, they cluster near the middle with the outlier well clear; lowest-to-highest crushes them into the bottom tenth.
Switching it on
It is the last card in the Score view's Advanced panel, under “How the factors combine” - deliberately, because that card decides how factors trade off against each other and this one only changes how each is measured on the way in.
Where to click
Map type
Score
Advanced
How each factor is placed
pick one of three
The card, set to 'How far from typical'. Each option's description says what it does to the map, not which statistic computes it.'Lowest to highest' is honest about its weakness in the description itself.
Same city, three maps
Three factors - cafés and transit stops within a walk, plus residents - three placements, nothing else changed. At city scale the three look more alike than you might expect, because the walk has already smoothed away the wildest outliers. Look at the edges of the bright core and at the outer boroughs, where the middling areas shift a shade between maps; the real differences live in individual areas, which is what the hovers in the next section show.
Where it ranks. A full spread of colour by construction: a third of areas are 'low', a third 'medium', a third 'high' on every factor.How far from typical. The bright core widens a little - many inner areas are genuinely above typical on all three - while ordinary outer areas settle toward the middle.Lowest to highest. Close to the ranked map here, because the walk tamed the outliers; with single-area counts instead, one café-packed block would drag the whole scale up and turn most of London dark.
One area, three scores
The same London area - three cafés, 1 891 residents, no transit stop, each read in just its own spot - under each placement. Nothing about the area changed. Its score moved by 65 points.
Where it ranks: 66 out of 100.How far from typical: 83 out of 100.Lowest to highest: 18 out of 100.
Why the scores move that much
Three cafés is a modest count, but London's café counts are wildly skewed: a few areas in Soho have dozens. Ranked, three cafés is above most areas. By distance from typical, it is a bit above typical. Lowest to highest, it is a tiny fraction of the maximum, so it scores near the floor. None of the three is wrong; each is answering a slightly different question about what “high” means.
Which one should you pick?
Exploring, and want to see the strongest and weakest areas at a glance? Where it ranks. It always fills the map with contrast and no outlier can break it.
Want genuinely exceptional places to stand out, and ordinary ones to look ordinary? How far from typical. It keeps the size of a gap without letting one area flatten the rest.
Data with no wild outliers, and you care about the true size of every gap? Lowest to highest. It is the most literal, and the most fragile.
The honest advice is the same as for the bivariate map's classification: try all three. If the big picture barely changes, your pattern is robust. If it changes a lot, a few areas sit on a boundary - which is worth knowing before you act on them.
Every factor is placed on a 0-1 scale before factors combine. Three honest ways: rank, distance from typical, lowest to highest.
Rank ignores the size of gaps and cannot be broken by an outlier. Lowest to highest keeps every gap and is broken by one.
Distance from typical is the middle path: gaps count, but three spread-widths cap the scale so one area cannot flatten the rest.
The same area scored 66, 83 and 18 under the three - not because it changed, but because 'high' means something different in each.
Flip between them and watch what moves. Robust patterns stay; boundary cases reveal themselves.
Quick questions
Why do factors need to be 'placed' at all?
Because you cannot add 14 cafés to 1 660 people. Each factor is first converted to a position between 0 and 1, and only then are the positions combined into a score.
What does 'Where it ranks' do?
It places each area by its rank against every other area: the median area sits at 0.5, the top at 1. One extreme area cannot stretch the scale, but the size of a gap is lost - 40 and 41 cafés look as different as 1 and 40.
What does 'How far from typical' do?
It measures how many spread-widths (standard deviations) an area sits from the typical value, so a big gap reads as a big gap. Three spread-widths either side of typical become 0 and 1, and anything beyond is capped.
What does 'Lowest to highest' do?
It places each area on the straight line from the smallest value to the largest. Gaps keep their size, but one extreme area stretches the whole scale and pushes everywhere else together.
Which one should I use?
Start with 'Where it ranks' for a full spread of colour. Switch to 'How far from typical' when you want genuinely exceptional places to stand out, and to 'Lowest to highest' only when your data has no wild outliers.
See it on a real map
The best way to understand the score is to play with it. The full app is on a free tier, with no sales call.
Builds MapBees's location-intelligence pipeline end to end — census and Overture Maps ingestion, H3 hex scoring, and the map interface — and writes the methodology behind every number on the site.
