| By Brian Livingston |
Despite the hotfix that Microsoft recently released for Excel 2007, as I described on Oct. 11, some math errors that you should know about still lurk in both Excel 2007 and Excel 2003.
I’ll bring you up to date and explain how you can get better results from Excel.
Baier and Neuwirth offer Excel math add-ins
In a nutshell, this month’s patch for Excel 2007 corrects a bug that treats numbers close to 65,535 as if they were 100,000. To get the fix, see the Oct. 9 entry in Microsoft’s official Excel blog.
Even with the hotfix, however, both Excel 2007 and Excel 2003 give slightly wrong — and, in some cases, extremely wrong — answers to some floating-point calculations. I’ll give you some examples below. First, let’s discuss an independent solution to the problem.
Those who want more accurate floating-point math than any version of Excel supports should download a statistics program called R. This is open-source software that was originally written by Robert Gentleman and Ross Ihaka (“R & R”), who now work with about 20 researchers around the world to maintain the code.
The R program, in turn, can be used with Excel if you install various add-ins by Thomas Baier and Erich Neuwirth called RExcel, rcom, and R(D)COM. Windows Secrets contributing editor Woody Leonhard recommended this in his Oct. 4 column on the Excel problem.