The trend analysis can show whether a trend can be identified when a series of measurements is viewed over the long term. Since trends usually develop very slowly due to wear, material aging or mechanical effects, it is necessary to use either very large amounts of data or suitable samples over a longer period of time.
Tip: When using very large data sets, it is advisable to use suitable import filters with only the relevant measurement IDs to keep the memory requirements small!
In the settings, the desired group is set first. In the Test Step Selection section, you can optionally hide entries of failed tests to avoid falsifying the result by bad parts. Finally, the system influence shall be recognized here.
In the section Floating Mean the window size for the floating average is set. A minimum of 20 or 10% of the total number is recommended (e.g. 100 for 1000 measurements)
Afterwards, the parameters whose trend is to be analyzed are selected. A curve is calculated for each value selected here and displayed in the "Floating Mean" panel. Note: The window may be in the background by default.
Calculate All: all available measurements are calculated
Calculate Selected: only the upper part of selected measurements are calculated
For each calculation, one result row is entered in the "Trend Analysis" table. When clicking on the respective row, the corresponding graphs are displayed.
History (Panel) This graph shows the course of the measured values (raw data) as well as a linear interpolation and an interpolation with a polynomial of adjustable degree. From this, an overall trend can already be read off if necessary.
Floating Mean (Panel) This graph shows the smoothed course of selected metrics of the measurement. The window size of the moving average can be adjusted.
Tip: If value ranges of the graphs do not fit together, deactivate individual graphs by clicking into the legend, if necessary, to achieve a better representation.