Visualisations

Last updated 2026-07-04

Visualisations turns your whole project's coding into charts — seven of them, all reading live from the same codebook and coded segments everything else in Paideias uses. Nothing here is edited or configured; you just pick a chart, and every one of them can be downloaded as a PNG or SVG for a report.

Project-wide, not per-document. Every chart on this screen aggregates across all your documents at once. If you want the coding for one specific document, this isn't that view — Retrieval and Focus Mode are where you work document-by-document; Visualisations is where you step back and see the shape of the whole project.

All seven chart types, in one recording

Below is a real recording: clicking through every tab in order, toggling the “Parent codes only” filter on the Treemap, and downloading a chart as a PNG — captured from a project with 44 codings across 18 codes and 2 documents.

paideias.org
Recorded, not staged
Every chart is one click away
Switching tabs re-renders instantly — there's no loading state, since every chart is computed from data already in memory. The sections below break down what each one shows and when to reach for it.

A stat strip above the tabs — Documents, Codes used, Total codings — stays visible no matter which chart is active, so you always have the raw numbers next to whatever you're looking at.

Code frequency

The default tab, and the one to start with. A horizontal bar chart of your top 16 codes by how many times each has been applied, sorted highest first.

Hierarchy

A two-ring sunburst: the inner ring is your parent codes, the outer ring is their child codes, both sized by total codings (a parent's inner slice includes its own direct codings plus everything coded under its children).

Treemap

Every used code as a rectangle, area proportional to coding frequency — a squarified layout, so even small codes stay visible instead of shrinking to a sliver.

Document × Code matrix

A heatmap: documents as rows, your top 12 codes as columns, each cell shaded by how many times that code appears in that document.

Connections

A bipartite diagram: documents on the left, codes on the right, curved lines connecting a document to every code it contains. Line thickness scales with how many times that code appears in that document.

Co-occurrence

Codes arranged in a circle, with a curved link between every pair of codes that ever appear in the same document. Thicker links mean that pair co-occurs more often.

Review status

The odd one out — instead of codebook structure, this donut shows where every AI suggestion in the project currently stands: approved, pending, or rejected.

Filtering to parent codes only

Four of the seven charts — Treemap, Doc × Code, Connections, and Co-occurrence — have a Show: toggle above them: All codes or Parent codes only. Switching it excludes every sub-code, showing only top-level codes and their own direct codings.

This filter only affects display, never your data. Toggling it changes what a chart shows, not what's coded — sub-code codings are always still there and still counted in “Total codings” and in every other screen.

Downloading a chart

⬇ PNG and ⬇ SVG in the top-right export whichever chart is currently on screen — not all seven at once, just the active tab. PNG renders at 2× scale for crisp printing; SVG stays fully vector, so it can be recolored or resized in design software without quality loss.

The filename always follows <project-name>_<chart-type>.png (or .svg) — for example, remote_team_trust_study_donut.png — so a folder of exports from the same project stays self-explanatory without opening each file.

Why these charts, not just the bar chart

Seven ways to look at the same underlying data might seem like overkill, but each chart genuinely answers a different question a bar chart can't:

ChartAnswers
Code frequencyWhich codes are used the most, overall?
HierarchyIs my codebook balanced, or top-heavy under one parent?
TreemapSame as frequency, but scannable at a glance without reading numbers
Doc × Code matrixIs a code concentrated in one document, or spread across all of them?
ConnectionsSame as the matrix, read as relationships instead of a grid
Co-occurrenceWhich themes tend to appear in the same documents together?
Review statusHow much AI-coding review work is left?

In practice, most write-ups only end up using two or three of these — frequency or treemap for a general overview, the matrix for cross-document comparison, and hierarchy if the codebook's parent/child structure is itself part of the argument. Having all seven costs nothing to skip past, and saves a re-export later when a reviewer asks for a different cut of the same data.

You're ready to write up your findings. With charts exported and passages pulled via Retrieval, the last step is Export & Download for a complete project archive, or Memos to capture the reasoning behind what you found.