DataLab is a compact statistics package aiming at exploratory data analysis. Please visit the DataLab Web site for more information....


Filter

Command: Math -> Filter...

Linear time-invariant filters (LTI filters) are a general class of filters which can be calculated by a simple equation. After clicking the command Math/Filter... a dialog is displayed which can be used to set up the filter parameters.

The data may be filtered either columnwise or rowwise. The columns/rows to be filered have to be selected after clicking the button Apply Filter. Depending on the selected storage option the filter results will be either appended to the matrix, or the original data will be replaced, or the results will be written into the matrix clipboard.

Besides a general FIR and IIR filter the filter dialog offers several commonly used standard filters. Following is a survey of the available options and settings:

Setting Description
Filter Type
  • General FIR Filter (FIR)
  • General IIR Filter(1) (IIR)
  • Moving Average (MA)
  • Weighted Average (WA)
  • Moving Median (MD)
  • 2nd Order Savitzky-Golay Polynomial (POL)
  • Smoothed 1st Derivative (1DV)
  • Smoothed 2nd Derivative (2DV)
No. of Coeff. N The number of filter coefficients may vary between 1 and 100. For polynomials and the smoothed 1st and 2nd derivative the number of coefficients must be odd and may vary between 5 and 25.
Load/Save Filter Coefficients Filter Coefficients which have been entered manually may be saved and loaded to and from the disk. The stored files are based on the ASC format, however, the file extensions are '*.FLT' (instead of '*.ASC') to improve operability.
Apply to Columns/Rows The user can choose to apply the filter function either to columns or to rows.
Application Range The application range specifies the range of columns/rows where the filter algorithm is applied.
Storage of Results The results of the filtering process can be either appended to the main matrix, or stored in the matrix clipboard, or they may replace the selected source data. The headings of the filtered data consist of an abbreviated identifier indicating the type of filtering and the length of the filter window followed by the original heading in parantheses. Example: the heading "F-WA7(C1)" indicates that column 1 has been filtered using a weighted average with 7 coefficients.



(1) The manual setup of the bi parameters of an IIR filter requires much know-how. In the case of unsuitable filter coefficents the results will be oscillating or will run towards infinity (leading to a floating point error). If you experience a floating point error when applying an IIR filter, you can be assured that this is due to bad filter coeffcients.