DataSmoother applies the current NeqSim Savitzky-Golay convolution to an equally spaced array. Use it for exploratory smoothing when the sampling grid is uniform and the chosen window is small relative to the signal features. Preserve the original measurements and their uncertainty: smoothing does not replace outlier assessment, uncertainty propagation, or an engineering data-quality review.

Current API

The public constructor is:

DataSmoother(double[] values, int leftPoints, int rightPoints, int derivativeOrder, int polynomialOrder)

The supported workflow is to construct the object with the input array, call runSmoothing(), and read the result with getSmoothedNumbers(). The class has no public no-argument constructor and no setInputNumbers method.

For the current implementation:

The implementation currently reports invalid parameter combinations rather than throwing a validation exception. Validate configuration before construction and do not treat a returned array from an invalid configuration as qualified output.

Executable Java 8 example

DataSmootherTest#documentationExampleCompilesAndRuns extracts this sole Java fence, compiles it with Java 8 source settings, enables assertions, and executes main. The numerical reference values are also protected by the existing direct unit test.

import neqsim.statistics.dataanalysis.datasmoothing.DataSmoother;

public final class DataSmootherGuideExample {
  private DataSmootherGuideExample() {}

  public static void main(String[] args) {
    double[] measurements = {
      10.0, 11.0, 12.0, 13.0, 14.0, 15.0, 15.5, 15.0, 19.0,
      14.0, 14.0, 13.0, 12.0, 12.0, 11.0, 10.0, 9.0, 8.0
    };

    int leftPoints = 3;
    int rightPoints = 3;
    int derivativeOrder = 0;
    int polynomialOrder = 4;
    DataSmoother smoother =
        new DataSmoother(
            measurements, leftPoints, rightPoints, derivativeOrder, polynomialOrder);

    smoother.runSmoothing();
    double[] smoothed = smoother.getSmoothedNumbers();

    assert smoothed.length == measurements.length;
    assert Math.abs(smoothed[0] - measurements[0]) < 1.0e-12;
    assert Math.abs(smoothed[3] - 12.567099567099579) < 1.0e-12;
    assert Math.abs(smoothed[8] - 16.66233766233768) < 1.0e-12;
    assert Math.abs(smoothed[17] - measurements[17]) < 1.0e-12;
  }
}

Interpretation and limits

The seven-point window in the example comprises three points on each side plus the center point. Polynomial order four controls the local least-squares basis; derivative order zero requests smoothing. The assertions protect two interior reference values and the current unchanged-edge behavior.

Do not interpret a visually smoother curve as evidence that the measurement is more accurate. Review window and polynomial sensitivity, compare with the raw data, retain units and sampling interval, and validate important trends against independent physical knowledge. Derivative modes require an explicit sampling interval scale and separate qualification; they are outside this example.