Filtering Data Q&A

What is the easiest way to smooth or denoise my data?

The Automated Smoothing option in the Filter menu is the easiest way to remove noise from the data. An AI Expert estimates the optimum smoothing level for each of the six automated procedures. Uniform data are automatically generated for procedures requiring constant X-spacing.

Why are there other procedures for smoothing and denoising?

For optimizing smoothing and denoising, three of the algorithms can be finely tuned in separate procedures. These require a constant X spacing:

The Savitzky-Golay Smoothing procedure offers effective time-domain smoothing for data sets with uniform X-spacing. The algorithm offers adjustable order, automatic sequential passes, and optional first through eighth smoothed derivatives. This is the recommended procedure for generating a smooth derivative.

The Fourier Denoising option is a specialized Fourier filtration procedure that sets either a frequency threshold for low pass frequency domain filtration, or a signal threshold for zeroing all spectral elements below a given power. The time domain data are reconstructed using the inverse FFT.

The Eigendecomposition Denoising option accomplishes a similar function except that the filtration occurs by zeroing those eigenmodes that contain noise. By using a high order decomposition, it is often possible to remove nearly all of the noise within a signal. This is often the best procedure for removing noise from a data set.

How do I isolate data components in the frequency domain?

You can manually edit the frequency representation of the data using the Fourier Filtering option in the Filter menu. This option offers real-time Fourier domain filtering and component isolation. This procedure supports exact-N FFTs and data tapering windows so that low power components can be isolated and reconstructed.

Apart from parametric fitting and Fourier filtering, can data components be isolated in any other way?

The Eigendecomposition Filtering option offers full eigenmode filtering and reconstruction. Eigendecomposition is a non-parametric procedure that partitions by signal strength rather than by frequency. In addition to the data, the reconstruction can optionally consist of the eigenvectors, the principal components, the data components, FFTs of the data components, or an FFT spectrum of the data.