hi peter, another question for you, if you are willing. well actually, both this question and the question i just asked Bert are for anybody willing to answer! I very much appreciate your opinions!
My dependent variable is an annual total. However most (tho not all) of the 250+ variables are at a daily resolution. I aggregated the x's up to annual levels, resulting in 17 years, n=17 because: 1) it seemed the data should be consistent, and reflect the limiting coarsest resolution - ... but maybe this is backward - maybe im losing too much information by aggregating? 2) this predictive model will be used by non-statasticians to experiment with how the Y will change when various Xs will change. Since the Y has to be at at an annual level, it seems weird to predict it by playing with changes in the Xs that take place on a daily level. For instance, one variable is the amount of pesticides used by stone fruit growers. I have totaled this to yearly amounts. hence a user of the model could experiment in changes of the Y if they thought pesticides likely to increase or decrease from previous years. If i 'un-aggragate' the data, i will have thousands upon thousands of data points - every single pesticide application reported in a year. however any model developed off of this will then be asking 'how will the annual Y change when an individual application of pesticide to stone fruit increases or decreases', and that question just doesnt really make any sense to me... I apologize in advance if i am in any way abusing this forum by posting so many questions on this topic. But I appreciate the help if you have any opinions! thanks -Kim -- View this message in context: http://r.789695.n4.nabble.com/regsubsets-Leaps-tp4632083p4632153.html Sent from the R help mailing list archive at Nabble.com. ______________________________________________ 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.
