www.phidot.org

Posted: **Fri Nov 02, 2012 10:39 am**

Hi,
Can anyone provide an example of how to generate predicted Nest Survival values in RMark for varying levels of a continuous covariate. There is a nice example in the RMark appendix of the MARK book of how to do this with varying levels of "Nest Age" and a continuous covariate, but I've had no luck applying the example to models that just include a continuous variable (not Nest Age). Here's what I've tried, but I'm pretty sure I have the structure of the design matrix all wrong. Any help would be appreciated.

Code: Select all: grsp.veg.height=mark(grsp,nocc=95,model="Nest",model.parameters=list(S=list(formula=~veg.height))) nocc=95 design=matrix(cbind(rep(1,(nocc-1)),seq(min(grsp$veg.height),max(grsp$veg.height),length=nocc-1)),nocc-1,2) get.real(grsp$veg.height,parameter="S",beta=grsp.vegheight$results$beta$estimate,se=T, design=design)

At this point, I should probably just use regular MARK, but I like working in R!

Thanks,
Mike

Posted: **Fri Nov 02, 2012 2:15 pm**

The mallard example was written before I created covariate.predictions which was designed to do what you want. You should be able to pass with arguments data=data.frame(veg.height=...) and indices=1, where ... is the set of values you want. I believe for your example that is correct but I may have overlooked something and if it doesn't work send me the data and R code and I'll look at it.

--jeff

Posted: **Fri Nov 02, 2012 3:02 pm**

Jeff,
I'll give covariate.predictions a try with the arguments you suggested. If I'm still stuck I'll send along the data.

Thanks,
Mike

Posted: **Wed Nov 07, 2012 11:31 am**

Hi Jeff,
That worked great - thanks! But one last (related) problem.

I'm now trying to graph both Year (2 dummy variables representing 2009, 2010, and 2012) AND a continuous covariate (% bare ground) simultaneously. It works well predicting DSR at various levels of the continuous variable ("bare"), but is ignoring my arguments for Year2009 (0 or 1) and Year2010 (0 or 1). It just returns the DSR as if both were 0, no matter what I put in. Here is the model and prediction code:

Code: Select all: #model gp1year.bare=mark(gp1,nocc=103,model="Nest",model.parameters=list(S=list(formula=~year2009+year2010+bare)), groups=c("year2009","year2010")) #beta estimates estimate se lcl ucl S:(Intercept) 3.0221740 0.5396361 1.9644872 4.0798608 S:year20091 0.3838193 0.6793767 -0.9477590 1.7153977 S:year20101 0.6545762 0.3754082 -0.0812239 1.3903763 S:bare -1.2772674 0.9206077 -3.0816586 0.5271237 #covariate predictions covariate.predictions(gp1year.bare,data=data.frame(bare=seq(min(gp1$bare),max(gp1$bare),length=10),year2009=rep(1,10),year2010=rep(0,10)),indices=1)

Again, no matter what value of Year2009 or Year2010 I put into covariate.predictions, it always returns the same set of dsr:

$estimates
vcv.index model.index par.index bare year2009 year2010 estimate se lcl ucl fixed
1 1 1 1 0.0000000 1 0 0.9535659 0.02389400 0.8770178 0.9833714
2 2 1 1 0.1230165 1 0 0.9460914 0.02231640 0.8815761 0.9764007
3 3 1 1 0.2460330 1 0 0.9374926 0.02007459 0.8845753 0.9670532
4 4 1 1 0.3690496 1 0 0.9276272 0.01762251 0.8845571 0.9554379
5 5 1 1 0.4920661 1 0 0.9163436 0.01645984 0.8779144 0.9434555
6 6 1 1 0.6150826 1 0 0.9034839 0.01929397 0.8584993 0.9352462
7 7 1 1 0.7380991 1 0 0.8888872 0.02748818 0.8225804 0.9324483
8 8 1 1 0.8611157 1 0 0.8723946 0.04038604 0.7705184 0.9329778
9 9 1 1 0.9841322 1 0 0.8538565 0.05737670 0.7034877 0.9350138
10 10 1 1 1.1071487 1 0 0.8331403 0.07825519 0.6235711 0.9376939

Any help is appreciated!

Thanks,
Mike

Posted: **Wed Nov 07, 2012 11:47 am**

Factor variables are handled with different real parameters. So you have to use the indices for those years to get different real estimates for them. Currently I suggested indices=1 because I didn't know you had differences for years. You can get those values from the design data or using the PIMS function with the simplified=FALSE argument.

--jeff

Posted: **Fri Nov 09, 2012 11:48 am**

Hi Jeff,
Thanks. I switched datasets on you and am kind of playing around as a learning experience. Below is what I got from the PIMS function. I'm still not clear on how to use that info to define the indices for each year. Is it something like this (since the index values for 2009 seem to be 103-204 based on PIMS)?:

Code: Select all: covariate.predictions(gp1year.bare,data=data.frame(bare=seq(min(gp1$bare),max(gp1$bare),length=10),year2009=rep(1,10),year2010=rep(0,10)),indices=103:204)

Sorry if this is too basic. I still haven't totally wrapped my head around how MARK handles data, design matrices, etc. behind the scenes, and have been heavily reliant on the example code.

Thanks again.
Mike

Code: Select all: > PIMS(gp1year.bare,simplified=F,parameter="S") group = year20090.year20100;stratum = 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 1 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 1 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 1 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 1 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 1 96 97 98 99 100 101 102 group = year20091.year20100;stratum = 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 1 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 1 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 1 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 1 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 1 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 96 97 98 99 100 101 102 1 198 199 200 201 202 203 204 group = year20090.year20101;stratum = 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 1 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 1 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 1 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 1 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 1 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 96 97 98 99 100 101 102 1 300 301 302 303 304 305 306

Posted: **Fri Nov 09, 2012 11:52 am**

No worries it is confusing. If 1 is an index to 2007 and 103 is an index to 2009, then you only need indices=c(1,103) to get the values for 2007 and 2009. You don't need to specify each real parameter unless you had differences based on the values of those real parameters.

--jeff

Posted: **Fri Nov 09, 2012 4:44 pm**

Thanks, that worked perfectly.

www.phidot.org

Predicted Nest Survival at Specified Covariate Values

Predicted Nest Survival at Specified Covariate Values

Re: Predicted Nest Survival at Specified Covariate Values

Re: Predicted Nest Survival at Specified Covariate Values

Re: Predicted Nest Survival at Specified Covariate Values

Re: Predicted Nest Survival at Specified Covariate Values

Re: Predicted Nest Survival at Specified Covariate Values

Re: Predicted Nest Survival at Specified Covariate Values

Re: Predicted Nest Survival at Specified Covariate Values