Hi,
This is a practical question and I am sure there are many statisticians can
give me a hand.
I have 500 time series data (500 rows), each row contains 100 intervals,
i.e., on each row, I have X1, X2, ..... X100. I am trying to reduce the
dimension of this input because the data at the end of each row does not
have significant meaning to the project I am doing.
I used cubic splines on ea. row -- ns(row, df = 10) , and decided to reduce
the dimension to 10. Then I generated the new feature by X_new = W*X, where
W is obtained from ns(row, df = 10). Now everything seemed perfect and I
was able to fit the logistic regression on the new inputs (500 rows, ea. row
has only 10 X's). After I computed the coefficients for this reduced model,
I want to map these coefficients back to the original problem. i.e., the new
coefficients has 10 betas, but I actually want 100 betas that I can use to
predict the response from the raw inputs. I donot know how to do this.
Any comments are highly appreciated.
thanks.
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
