| Summary: | Data = Smooth (as Fit) + Residuals | ||
|---|---|---|---|
| Product: | [Applications] LabPlot2 | Reporter: | disuser |
| Component: | general | Assignee: | Alexander Semke <alexander.semke> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | ||
| Priority: | NOR | ||
| Version First Reported In: | latest | ||
| Target Milestone: | --- | ||
| Platform: | Other | ||
| OS: | Linux | ||
| Latest Commit: | Version Fixed In: | 2.8 | |
| Sentry Crash Report: | |||
|
Description
disuser
2020-04-25 06:47:56 UTC
The primary reason for examining any residual plot is to see whether the rough is rough enough or whether it contains additional smooth. Data = Fit + Residual Data = Smooth + Rough As John Tukey explained: "Here the fit is our current description --always incomplete, always approximate--of the overall behavior of the data. Each individual observation is split up into a sum of this fit and what is left over, called a residual. Residuals are our main tool in going further. They are to the data analyst what powerful magnifying glasses, sensitive chemical tests for bloodstains, and delicate listening devices are to a story-book detective. They permeate all sorts of data analysis and appear in many guises. [...] For us the most useful plot will be one that might reveal the unexpected or the unobvious. Sometimes a plot "in the large" will do this. Usually, however, it is the plot of residuals that has the greatest use and the greatest impact." |