Curve Fitting API

The Curve Fitting system allows you to mathematically analyze data series and visualize trends. It uses high-performance numerical solvers to compute regressions and automatically renders them as stylized fit lines with equations.

Requirement

The Curve Fitting system is part of the PluginAnalysis module. You must load this plugin to use addFitLine.

import { createChart } from 'scichart-engine';
import { PluginAnalysis, PluginAnnotations } from 'scichart-engine/plugins';

const chart = createChart({ container });

// Enable analysis and annotations (for equation labels)
await chart.use(PluginAnalysis());
await chart.use(PluginAnnotations());

Using addFitLine

The most common way to use curve fitting is via the addFitLine method on the Chart instance.

Automatic Annotations

When you call addFitLine, the engine calculates the regression, adds a new themed dashed series for the fit, and automatically adds a text annotation containing the equation and $R^2$ value. This requires PluginAnnotations to be loaded.

Async Safe

You can call addFitLine immediately after creating the chart. If the analysis plugin hasn't finished loading yet, the request will be queued and applied automatically as soon as the plugin is ready.

Signature

chart.addFitLine(seriesId: string, type: FitType, options?: FitOptions): string;

Parameters

Name	Type	Description
seriesId	string	The ID of the series to analyze.
type	FitType	The regression model to use.
options	FitOptions	Optional configuration for the fit.

Returns

• string: The ID of the newly created fit series.

Supported Models (FitType)

Type	equation	Use Case
'linear'	$y = ax + b$	General trends, calibration curves.
'polynomial'	$y = a_n x^n + ... + a_0$	Non-linear trends, complex curves.
'exponential'	$y = a \cdot e^{bx}$	Growth, decay, kinetics.
'logarithmic'	$y = a + b \cdot \ln(x)$	Signal saturation, pH changes.
'power'	$y = a \cdot x^b$	Scaling laws, physics models.

Options (FitOptions)

interface FitOptions {
  /** Degree for polynomial fit (default: 2) */
  degree?: number;
  /** Number of decimals in equation string (default: 4) */
  precision?: number;
  /** Custom label for the equation annotation */
  label?: string;
}

Low-level API (fitData)

If you want to perform the regression math without automatically adding a series to the chart, you can use the fitData utility.

import { fitData } from 'scichart-engine/analysis';

const result = fitData(xValues, yValues, 'polynomial', { degree: 3 });

console.log(result.equation); // "y = 1.25x^3 - 0.50x^2 + ..."
console.log(result.rSquared); // 0.9985
const predictedY = result.predict(2.5);

FitResult structure

interface FitResult {
  type: FitType;
  coefficients: number[];
  equation: string;
  rSquared: number;
  predict: (x: number) => number;
}

Mathematical Implementation

The engine uses Gaussian elimination with partial pivoting to solve the normal equations of the least squares problem. This provides robustness against ill-conditioned matrices common in high-degree polynomial fitting.

For non-linear models (Exponential, Power, Logarithmic), the data is linearized using logarithmic transformations before performing a linear regression.