1

I have a matrix with 11 columns named "env", with 1 response variable bounded between 0 and 1 ("R1") and 10 possible predictors ranging from "P1 to "P10". I would like to use a zero-inflated beta regression (r package and function "gamlss") to assess the individual effect of each predictor on my response variable summarizing the AIC, estimate and probability for each predictor in a table. The table should have the predictors as rows and model parameters (AIC, estimate and probability) as columns. This process must be repeated individually for the three coefficients of the beta distribution (MU, NU and SIGMA).

Here a subset of my data matrix (sorry for not being able to simulate it following the guidelines)

P1  P2  P3  P4  P5  P6  P7  P8  P9  P10 R1
600 243.89  2.68    180.32  1753.62 5.15    16.11   46.59   1.52    0.96    0.04
674 259.43  1.49    174.06  1230.71 5.50    19.42   45.65   1.62    0.88    0.28
231 156.19  0.00    151.68  1002.93 5.22    12.76   50.38   1.63    1.00    0.00
624 256.53  8.58    181.32  1194.07 5.35    25.33   58.74   1.33    0.94    0.36
773 346.91  15.59   180.17  1665.10 4.99    26.65   39.74   1.13    0.93    0.21
468 186.84  6.13    172.11  1570.75 5.34    18.72   38.52   1.55    0.97    0.10
340 478.28  14.68   169.06  1685.20 4.81    14.17   112.48  1.65    0.98    0.00
719 401.34  14.57   180.84  1824.18 4.74    13.46   129.70  1.67    0.98    0.00
603 831.58  7.79    158.69  1675.99 5.49    35.08   109.76  1.40    0.62    0.00
355 463.96  3.39    174.65  1987.08 4.26    25.69   85.57   1.56    0.98    0.03
527 560.11  32.18   175.29  2661.40 4.69    50.79   84.67   1.30    0.92    0.14
603 313.94  20.98   163.86  3211.07 4.60    45.15   86.36   1.36    0.93    0.02
508 571.58  40.62   118.69  2842.65 5.11    53.89   57.88   1.31    0.99    0.13
270 191.50  0.35    176.33  3280.57 4.75    51.99   127.10  1.29    0.51    0.12
353 770.72  0.05    173.76  2079.63 5.46    39.12   141.40  1.26    0.51    0.16
166 488.43  12.40   164.20  2692.61 4.55    41.28   107.06  1.40    0.91    0.13
881 316.41  32.43   156.37  2883.55 4.15    29.20   71.32   1.59    0.89    0.21
013 734.83  20.08   156.98  2044.81 4.72    49.62   113.42  1.35    0.98    0.20
526 452.85  33.85   164.58  1795.64 5.01    26.16   87.38   1.54    0.95    0.06

Below the syntax for the gamlss function with the 3 coefficients:

m1.mu<-gamlss(formula= R1~P1, family=BEZI, data=env, method=RS(100))
m1.nu<-gamlss(formula= R1~P1, nu.formula= R1~P1, family=BEZI, data=env, method=RS(100)) 
m1.si<-gamlss(formula= R1~P1,si.formula= R1~P1, family=BEZI, data=env, method=RS(100))

and here is the structure of the table I am trying to get:

    AIC.mu  EST.mu  PROB.mu AIC.nu  EST.nu  PROB.nu AIC.si  EST.si  PROB.si
P1                                  
P2                                  
P3                                  
 …                                  
P10

I assume I should be able to automate this with a combination of "for loop", "lapply" or "apply" and "cbind" but unfortunately I cannot manage to get it working. Would be great if some of you guys could give me a hand. Many Thanks

R Trancoso
  • 35
  • 4
  • Welcome to SO. Please carefully read [this](http://stackoverflow.com/help/mcve) and [this](http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example/5963610#5963610) and then update your question to conform to these guidelines. In particular, no one can run your code unless we have `env`. – jlhoward Sep 30 '15 at 06:48
  • Fair enough - thanks for your handy hints for a SO rookie. – R Trancoso Sep 30 '15 at 07:22
  • Are you able to extract what you need from one set of models? – Heroka Sep 30 '15 at 10:01
  • Thanks for your reply Heroka. Yes I am. They are all displayed using the "summary" function. I can use m1.mu$mu.coefficients and m1.mu$aic to get the estimates and AIC (for the first model for instance), but I don't know how to get the probability using this syntax as well as to collect them all and report in a summary table. – R Trancoso Oct 01 '15 at 03:15

0 Answers0