VIM for depthmapX

Visibility Integration Mismatch shows where a plan is unintelligible. Each point of a visibility graph is compared with points that have a similar-sized view: hidden centres are more integrated than their view suggests, false centres look open but sit at the edge of the layout.

Runs entirely in your browser. Your file is never uploaded.

How to export from depthmapX

Import your plan, Set Grid and fill the spaces.
Tools → Visibility → Make Visibility Graph.
Tools → Visibility → Run Visibility Graph Analysis: tick Calculate visibility relationships and Include global measures (radius n).
With the visibility graph displayed, File → Export and save as .csv or .txt. Also save your graph (File → Save) if you want to put VIM back onto the map later.

Works with any depthmapX version whose export has Connectivity and Visual Integration [HH] columns.

Drop your depthmapX export here or click to choose a .csv or .txt file

Options

Connectivity bands Min. points per band Strong hidden centre: VIM ≥ Moderate hidden centre: VIM ≥ False centre: VIM ≤ Plan units (for the scale bar)

The defaults are the published method. Change them only for sensitivity analysis; the summary records the values used.

Map

Number the top clusters

Downloads

The map PNG is the current view at print resolution, with title, legend and scale bar. The results table is your export with the six VIM columns added.

Put VIM back onto your map in depthmapX

The join file holds Ref and the VIM columns. depthmapX's command-line tool can attach it to the visibility graph in your saved .graph:

depthmapXcli -m IMPORT -f plan.graph -o plan_vim.graph -if plan_vga_vim_join.csv -iaa

Open plan_vim.graph in depthmapX: the VIM columns are in the attribute list. Colour by VIM Map (VIM+5) or VIM Class; depthmapX shows negative numbers as "no value", so plain VIM leaves false centres blank.

Classes

Correlations (Pearson r)

Top hidden-centre clusters

Method

Valid points have connectivity ≥ 0 and visual integration ≥ 0 (Visual Integration [HH]).
Bands. Points are ranked by connectivity and split into 40 equal-sized quantile bands. Points with identical connectivity always share a band (the tie group goes to the band of its middle rank). Small plans get ⌊n/10⌋ bands so each keeps at least ~10 points.
Expected integration in each band: the median and the sample standard deviation.
VIM = (integration − band median) ÷ band SD; 0 when the SD is 0.
Classes: strong hidden centre VIM ≥ 2 · moderate hidden centre 1 ≤ VIM < 2 · false centre VIM ≤ −1 · otherwise as expected.
Clusters: strong hidden-centre cells that touch, diagonals included, form a cluster. Clusters are ranked by summed VIM, so a cluster ranks high by being strong, large or both.

Because each point is compared only within its connectivity band, VIM is independent of visual-field size by construction. This page gives the same values as the depthmapx_vim.py command-line tool and the C++ implementation, to depthmapX's export precision.