Embed in your application. Flag outliers in your business data. Add value to your users. Add client-side anomaly detection to any frontend app in minutes.
This isn't a screenshot. Switch the algorithm or change the window size in the toolbar and the chart re-scans live using the WebAssembly engine.
Integrate into any framework
One package, four drop-in entry points. Install@datastripes/anomaly once, import the entry point for your stack, and render anomaly detection with a single component.npm install @datastripes/anomaly
import { DatastripesAnomaly } from '@datastripes/anomaly/react';
export function App() {
return <DatastripesAnomaly data={myRows} algorithm="ensemble" />;
}One engine, real detection
The core engine does the actual work; the UI components are a thin layer on top.01
Auto column discovery
Hand it an array of records, a CSV string, or raw JSON and it infers the datetime column and the target metric by name and type, no manual schema required. For scatter data, it picks the two best numeric columns for an (x, y) pair instead.02
Six detection algorithms
Global z-score, rolling (windowed) z-score, IQR, MAD, STL-lite residual decomposition, and an ensemble majority vote, each with tunable thresholds and a reported anomaly count, rate, mean, and max score.03
Severity & method comparison
Every flagged anomaly is scored low/medium/high severity, and a side-by-side leaderboard shows how many anomalies each method finds and how much they agree with each other, all computed in WebAssembly in the browser.Fully customizable
Every surface is configurable through typedcustomOptions, chart colors, KPI cards, and diagnostics.04
Chart styling
ChartStyleOptions controls per-series colors, baseline band colors, grid, tooltips, y-axis formatting, and a renderMode toggle between a connected line and a per-point bar chart.05
KPI cards & toolbar
KpiCardsOptions picks which metrics appear; HeaderToolbarOptions toggles the algorithm selector, window-size input, and export buttons.06
Views, tabs & export
Switch between composite, chart, or table views. Configure leaderboard tabs and control which export formats (CSV, JSON, PDF) are offered.Built for developers
What users get
✓Interactive anomaly exploration with pan and zoom
✓Anomalies marked by severity, in line or bar render mode
✓A KPI deck summarizing anomaly count and rate at a glance
✓Side-by-side method comparison across all six algorithms
✓A column mapping modal to review auto-discovered columns
What developers get
✓One npm package with subpath entry points
✓A framework-agnostic core entry with zero dependencies
✓Fully typed API: props and options are TypeScript interfaces
✓Safe-by-default licensing with a restricted trial tier
✓WASM-powered computation off the main thread
Start free, upgrade when you need unlimited rows, exports, and no watermark.