Grandscatter

Interactive multidimensional scatterplot widget for Jupyter notebooks. Rotate projection axes by dragging to explore high-dimensional point clouds with correct linear projections at all times.

Built on anywidget and WebGL.

Installation

pip install grandscatter

Quick start

from grandscatter import Scatter
import pandas as pd

df = pd.read_csv("my_data.csv")

widget = Scatter(
    df,
    axis_fields=["x1", "x2", "x3", "x4", "x5"],
    label_field="category",
    label_colors={"A": "#e23838", "B": "#2196f3", "C": "#4caf50"},
)
widget

Features

• Interactive axis rotation -- drag axis handles to rotate the projection and explore your data from any angle.
• Orthogonal projections -- the projection matrix is always kept orthonormal, ensuring geometrically correct linear projections.
• Perspective and orthographic modes -- switch between projection types on the fly.
• WebGL rendering -- fast, anti-aliased point rendering with depth sorting.
• Categorical legend -- click legend items to highlight categories.
• Live trait sync -- update properties like projection, axis_length, view_angle, and base_point_size from Python and see changes reflected immediately.

API

Scatter(df, axis_fields, label_field, label_colors, **kwargs)

All keyword parameters are traitlets and can be updated after creation:

widget.projection = "perspective"
widget.base_point_size = 4
widget.view_angle = 90

License

MIT