| This seems like a perfectly appropriate question...
I'm not sure about all of these terms, but "factor analysis" is a statistical
technique for analyzing experimental data to understand what inputs or
combination of inputs in a series of experiments had the most effect on the
result, and some factor analysis techniques use rotation of eigenvectors. The
other terms also sound like statistical analysis terms but I don't know them.
I don't know of any PC statistical analysis packages, although I'm sure there
are some. The non-PC package I remember that does this kind of stuff is SPSS.
If your daughter has access to university faculty, she can try the experimental
sociology or psychology department if the chem department doesn't have a
statistical analysis package; those fields do lots of statistical analysis.
|
| The field of mathematics actually goes under the names of
"Principal Components Analysis" and "Factor Analysis",
the idea being to look for multidimensional correlation between
data points. Eigenvalue and singular value decomposition
routines are tools used by these packages. There are lots
of packages out there and many books with titles ivolving
keywords like "principal components". I think SPSS does
this kind of thing.
For MATLAB, there is an email address and they'll know
what to recommend. The big problem is not the core calculations
for which solid routines are publically available and well known,
but all the data capture, graphical display of results, etc.
One place to look for other references is "Numerical Recipes"
which gives an overview of lots of useful ideas and insights about
numerical computations.
I think the term "Factor Analysis" was invented by Hotelling
in the 30's.
The big problem in my opinion with all this is that everything
is assumed linear with gaussian noise, not all that realistic
in practice... but there is some nice linear algebra math
involved.
- Jim
|