https://issues.apache.org/ooo/show_bug.cgi?id=123824
Bug ID: 123824
Issue Type: DEFECT
Summary: Exponential curve fit y=a*exp(b*x) ignores data points
with zero y-values
Product: General
Version: 4.0.1
Hardware: PC
OS: Windows 7
Status: UNCONFIRMED
Severity: critical
Priority: P3
Component: chart
Assignee: [email protected]
Reporter: [email protected]
Created attachment 82063
--> https://issues.apache.org/ooo/attachment.cgi?id=82063&action=edit
spreadsheet with charts to demonstrate y=exp(x) problem
When choosing an exponential trend line y=a*exp(b*x) through some actual data I
realized that a number of the data point that had zero y-values were being
ignored. This is different from a reported bug that described a curve fit not
going through a data point at the origin. Although no exponential regression
of the form y=a*exp(b*x) can intersect the x-axis, there should be no reason
why a minimization of least square values between data and candidate
regressions should have a problem. The attached spread sheet demonstrates the
problem. I did not try to demonstrate what a data point within the x-range of
data with a nonzero y-value would do, but the example shown is enough to
demonstrate a problem. Note the example was created and showed the problem
before it was saved in .xls format for the first time for whatever that is
worth.
--
You are receiving this mail because:
You are the assignee for the bug.
You are watching all bug changes.
