Pavlos,
There are several ways to evaluate how well new data fit an old
regression.Part of the answer depends on what you are concerned about. For
example, if you are concerned about bias, you can test whether the mean of
the new data is within the expected range of the mean of that many new
values. The equations for these prediction intervals should be in good
texts on linear regression.
Dave
Date: Sun, 27 Oct 2013 13:36:12 +0200
From: Pavlos Pavlidis <pavlidisp@gmail.com>
To: r-help <r-help@r-project.org>
Subject: Re: [R] how well *new data* fit a pre-computed model
Message-ID:
<CABZ9MBUEuCVbF0WKRFuBtbWYKwjYzvvJmusmrypTFPa+yB=QSQ@mail.gmail.com>
Content-Type: text/plain
Here is a link to a plot that illustrates the question:
bio.lmu.de/~pavlidis/pg1.pdf
the question is how to evaluate whether the blue points fit the curve well
enough. The curve has been produced from the black points
best
pavlos
On Sun, Oct 27, 2013 at 1:30 PM, Pavlos Pavlidis
<pavlidisp@gmail.com>wrote:
> Hi all,
> I have fitted polynomial models to a bunch of data (time-course analysis).
> The experiment is "the expression value of gene A under condition K
over
> time". The data points that have been used to fit the model are about
200
> (dataset A). Furthermore I have a few data (dataset B; about 10 points)
for> "the expression values of gene A under condition G over time".
The
question> is:
>
> how can I evaluate how well the dataset B fits the model generated by
> dataset A?
>
> kind regards,
> pavlos
>
> --
>
> Pavlos Pavlidis, PhD
>
> Foundation for Research and Technology - Hellas
> Institute of Molecular Biology and Biotechnology
> Íikolaou Plastira 100, Vassilika Vouton
> GR - 711 10, Heraklion, Crete, Greece
>
--
Pavlos Pavlidis, PhD
Foundation for Research and Technology - Hellas
Institute of Molecular Biology and Biotechnology
Íikolaou Plastira 100, Vassilika Vouton
GR - 711 10, Heraklion, Crete, Greece
[[alternative HTML version deleted]]
------------------------------
[[alternative HTML version deleted]]