Charts

Chart types

The 29 chart types, what each is for and how to add one.

Kymoscope draws 29 kinds of chart from the same datasets. Every chart in these pages is drawn live by the engine from the data of its example, so what you copy is what you see.

Add a chart

A chart is a kind plus settings. Settings name columns of the datasets in the worksheet; the chart draws every selected dataset, and shows an empty state until its columns are set.

TypeScript
import { createEngine } from '@kymoscope/sdk'

const engine = createEngine({ licenseKey })
await engine.datasets.addText(await file.text(), file.name)

const chart = engine.charts.add('histogram', {
  title: 'Shaft diameter',
  settings: { histColumn: 'diameter', histNormal: true },
})
engine.charts.mount(chart, document.getElementById('chart')!)

// Later: change any setting; the kind stays.
engine.charts.update(chart, { histBins: 40 })

Settings you leave out take their defaults (listed under each chart). In the app, the same settings are in the chart's settings dialog. See Worksheets and charts for layout and windows, and Mounting and styling charts for colours and themes.

Which chart to use

ChartUse it for
Multi-bandThe main time-series chart.
Area ChartFilled areas of several columns over an X column, stacked to show how parts add up to a total (energy sources, load per line, queue by state).
SpectrumThe FFT of one signal.
SpectrogramA short-time FFT.
Control ChartStatistical process control.
GaugesDials and bars for the current value of a few channels.
CameraVideo of a camera next to its data.
X-Y scatterOne column against another, to see how they relate.
Bubble ChartA scatter where each point is a bubble sized (and optionally coloured) by more columns.
X-Y-Z meshThree columns in 3-D.
HeatmapA value as colour over a grid.
PolarA value against an angle.
Parallel CoordsMany columns at once.
RadarA profile across several measures, one polygon per group.
HistogramHow the values of one column are distributed.
Box PlotThe distribution of a value per group as boxes (quartiles, median, whiskers, outliers).
ViolinLike a box plot, with the full shape of each distribution (kernel density).
CDF / ECDFThe empirical cumulative distribution.
Q-Q PlotThe quantiles of a column against a normal distribution (or against another column).
Bar ChartValues per category as bars, grouped or stacked.
TreemapA hierarchy as nested rectangles sized by value.
SankeyFlows between stages as bands whose width is the amount.
Flame GraphA hierarchy as stacked bars whose width is the value.
Gantt ChartIntervals on a time axis, one bar per row.
Confusion MatrixActual against predicted classes of a classifier.
ROC CurveHow well a score separates two classes at every threshold.
Calibration CurveWhether predicted probabilities mean what they say.
Loss CurvesTraining curves on two Y axes.
EmbeddingA 2-D embedding (t-SNE, UMAP, PCA output) coloured by class or by value.

By family

What every chart shares

  • Reference lines. Most kinds take an alarm line (histAlarm, bcAlarm, …): { value, style, width, color }, or null for none. A line linked to an alert rule follows the rule's threshold (see Alerts).
  • Annotations. Text, arrows, boxes and strokes drawn on any chart, saved with it (see Worksheets and charts).
  • Cursor and zoom. Charts of a worksheet share its X range and cursor; multi-band charts set the cursor on hover.
  • Theme. Every chart follows dark or light mode; mount options set its colours.
  • Text columns. Category, name and class columns can hold text when the data comes from CSV (datasets.addText or an imported file); addColumns takes numbers only. A units row under the header is optional; when the reader guesses the rows wrong, set parseConfig with datasets.configure.
  • Several datasets. Selected datasets draw together, each in its own colour, so runs compare in the same chart.