Visualisation

Overview

core.visualisation provides linked scatter, histogram and density views for the active dataset. It combines HoloViews and Datashader rendering with bounded interactive data access so large datasets can still participate in focus and selection workflows.

What You Do

  1. Open a standalone plot or the combined Linked Plot Explorer.

  2. Choose the X and Y columns and, where useful, a colour or grouping column.

  3. Inspect numeric, datetime or categorical structure in the active dataset.

  4. Tap a scatter point to focus one record.

  5. Use box or lasso selection in the scatter view to create a selection set.

  6. Open Record Browser, Image, Spectra, SED or Selection Tools to inspect the same focus and selection.

  7. Save the workspace to restore the chosen plot settings later.

Required Mappings

  • record_id

No label mapping is required. If a target_label mapping exists it can be used as the default colour or grouping column, but any supported dataset column can be selected instead.

Panels

Scatter Plot

The Scatter Plot supports configurable X and Y columns, optional categorical or continuous colouring, sampled interactive rendering and Datashader fallback for large datasets.

Tapping a point changes focus. Box and lasso tools create selection sets, with the maximum number of published row IDs controlled by the panel settings.

../../_images/scatterplot_example_short.gif

Histogram Plot

The Histogram Plot displays the selected X variable and can split the distribution by a categorical colour column. It supports configurable bins, value limits, percentage or cumulative display, logarithmic axes and an overlay for the focused row.

Categorical X values are represented as discrete histogram bins.

Density Plot

The Density Plot provides a two-dimensional density view for the selected X and Y columns. It supports interactive or rasterised rendering, configurable density bins, axis limits and logarithmic axes.

The focused row can be shown on top of the density result without turning every source row into an interactive point.

../../_images/density_example_short.gif

Linked Plot Explorer

The Linked Plot Explorer places scatter, histogram and density views in tabs behind one shared settings header. Axis, colour, filtering and rendering changes are shared between its three child views.

Standalone plot instances use their own visualisation state and remain independent of one another.

Focus and Selection

Focus and selection are separate platform states. A scatter tap normally changes focus, which identifies one record. Box or lasso interaction creates a selection, which contains a set of record IDs.

The plugin uses the platform SelectionManager rather than directly updating other panels, so linked plugins can react to the same focus and selection.

Large Datasets

Scatter and density views can switch to Datashader when the dataset exceeds the configured rendering threshold. Interactive scatter mode uses a bounded sample for browser-level point interaction while retaining the larger-data context.

Shared prepared-frame, column, row-ID and focused-row caches reduce repeated source reads across related visualisations. Focused rows can also be resolved asynchronously through the platform job system when a direct lookup would be too expensive for the UI thread.

Axis Types

Plot axes support scalar numeric, datetime and categorical values. Numeric-like and datetime-like strings are converted when they can be interpreted reliably; remaining scalar values are factor encoded for categorical display.

Array, struct, map, binary and similar non-scalar columns are excluded from the normal axis choices.

Saved State

Axis choices, colour and filtering settings, render mode, sampling limits, point appearance, histogram and density settings and supported axis limits are stored with each panel instance.

The Linked Plot Explorer stores one shared state for its child views. Standalone plot panels restore their own state independently.

Note

Rasterised pixels describe aggregate density rather than individually selectable records. Use the scatter interaction layer when precise focus or row-set selection is required.