https://communities.sas.com/t5/Statistical-Procedures/Environmental-Time-Series-de-tangling-co-variance/m-p/65819#M3139 I'm a fisheries biologist and i'm scouting around for a robust way to determine what the principle environmental influences are on a number of response variables such as fish swimming speed or travel time between two points.<BR /> <BR /> The environmental variables are daily mean water temperature, daily precipitation, barometric pressure, and river flow rate.<BR /> <BR /> The challenge is that these variables are linked -- large rainfalls drop the water temperature and increase the river flow rate; high barometric pressure is associated with high water temperatures and low flow rates.<BR /> <BR /> Another challenge is that most of the environmental variables are not normally-distributed, so parametric statistics may not be suitable without specific transformations. Are there recommended transformations for rainfall data, for example?<BR /> <BR /> Finally, there is the temporal nature of the data -- with auto-correlation in all variates. <BR /> <BR /> Is there a statistically-defensible process to decipher which predictor or predictors are key?<BR /> <BR /> Thanks! Fri, 13 May 2011 17:07:13 GMT WaterColour 2011-05-13T17:07:13Z https://communities.sas.com/t5/Statistical-Procedures/Environmental-Time-Series-de-tangling-co-variance/m-p/65819#M3139 I'm a fisheries biologist and i'm scouting around for a robust way to determine what the principle environmental influences are on a number of response variables such as fish swimming speed or travel time between two points.<BR /> <BR /> The environmental variables are daily mean water temperature, daily precipitation, barometric pressure, and river flow rate.<BR /> <BR /> The challenge is that these variables are linked -- large rainfalls drop the water temperature and increase the river flow rate; high barometric pressure is associated with high water temperatures and low flow rates.<BR /> <BR /> Another challenge is that most of the environmental variables are not normally-distributed, so parametric statistics may not be suitable without specific transformations. Are there recommended transformations for rainfall data, for example?<BR /> <BR /> Finally, there is the temporal nature of the data -- with auto-correlation in all variates. <BR /> <BR /> Is there a statistically-defensible process to decipher which predictor or predictors are key?<BR /> <BR /> Thanks! Fri, 13 May 2011 17:07:13 GMT https://communities.sas.com/t5/Statistical-Procedures/Environmental-Time-Series-de-tangling-co-variance/m-p/65819#M3139 WaterColour 2011-05-13T17:07:13Z https://communities.sas.com/t5/Statistical-Procedures/Environmental-Time-Series-de-tangling-co-variance/m-p/65820#M3140 I am sorry, but your questions are much too general to be addressed here. You are basically asking for an entire course in empirical modeling. I suggest you find a consulting statistician who can work with you on this problem. Mon, 16 May 2011 12:44:13 GMT https://communities.sas.com/t5/Statistical-Procedures/Environmental-Time-Series-de-tangling-co-variance/m-p/65820#M3140 lvm 2011-05-16T12:44:13Z