<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <div class="moz-cite-prefix">Dear Jesus,<br>
      <br>
      I agree with your comment, that's why we don't want to give a mean
      or a median of variables importance of all single models selected
      as variable importance... <br>
      <br>
      As for ensemble models evaluation, we have to consider ensemble
      models as "true" models.. So we have to apply the same methodology
      than for the single models..<br>
      We must randomise a variable, calculate all single models
      predictions with this 'new set of data', combine predictions to
      build our ensemble model, calculate the correlation between the
      'true' predictions of our 'ensemble model' and the ones with a
      randomised variable....Do it several times for all variables and
      all models.... It will be really time consuming but might be
      interesting in some cases.<br>
      <br>
      I guess that calculating variables importance according to
      methodology describes above will give you a quite interpretable
      value... <br>
      <br>
      Cheers,<br>
      <br>
      Damien.<br>
      <br>
      <br>
      On 13/09/2012 09:57, Jesus wrote:<br>
    </div>
    <blockquote
cite="mid:CA+by4T77VnrTCVq7XsTtkeSteHrtn6buWgdpo5HFHcUhEdauww@mail.gmail.com"
      type="cite"><br>
      <br>
      <div class="gmail_quote">On 13 September 2012 09:49, Damien
        Georges <span dir="ltr"><<a moz-do-not-send="true"
            href="mailto:damien.georges2@gmail.com" target="_blank">damien.georges2@gmail.com</a>></span>
        wrote:<br>
        <blockquote class="gmail_quote" style="margin:0 0 0
          .8ex;border-left:1px #ccc solid;padding-left:1ex">
          <div bgcolor="#FFFFFF" text="#000000">
            <div>Dear Pablo,<br>
              <br>
              There is no way to extract variables importance from a
              Ensemble model because it's not calculated yet.. We are
              thinking about the way to implement it  because it's a
              really time consuming step.. Concerning the response
              plots, the problem is the same.. <br>
              I will tell you when this functionalities will be ready.<br>
              <br>
              Cheers,<br>
              <br>
              Damien. <br>
              <div>
                <div class="h5">Dear Damien, thanks for the responses to
                  this threat. I however have a question/observation
                  about the variables importance in a ensemble of
                  models.  How do you really interpret the variable
                  importance coming from an ensemble of models? Imaging
                  two algorithms gave you high AUC values so they are
                  considered good models, but choose different
                  environmental variables, what would be the variable
                  importance result of the ensemble? I find the
                  interpretation of this a bit tricky. and not that
                  straight forward.</div>
              </div>
            </div>
          </div>
        </blockquote>
        <div><br>
        </div>
        <div>Jesus </div>
        <blockquote class="gmail_quote" style="margin:0 0 0
          .8ex;border-left:1px #ccc solid;padding-left:1ex">
          <div bgcolor="#FFFFFF" text="#000000">
            <div>
              <div>
                <div class="h5">
                   </div>
              </div>
            </div>
          </div>
        </blockquote>
        <div> </div>
        <blockquote class="gmail_quote" style="margin:0 0 0
          .8ex;border-left:1px #ccc solid;padding-left:1ex">
          <div bgcolor="#FFFFFF" text="#000000">
            <div>
              <div>
                <div class="h5">On 12/09/2012 17:59, Pablo Riul wrote:<br>
                </div>
              </div>
            </div>
            <div>
              <div class="h5">
                <blockquote type="cite">
                  <div style="font-size:18pt;font-family:times new
                    roman,new york,times,serif">
                    <div><span>Dear Damien,</span></div>
                    <div
                      style="font-style:normal;font-size:24px;background-color:transparent;font-family:'times
                      new roman','new york',times,serif"><span><br>
                      </span></div>
                    <div
                      style="font-style:normal;font-size:24px;background-color:transparent;font-family:'times
                      new roman','new york',times,serif"><span>Ok,
                        thanks! Just another thing... I'm not sure if
                        you received my anterior message. </span> I
                      tried to get the var importances of a consensus
                      model using something like getModelsVarImport() to
                      an object created by the BIOMOD_EnsembleModeling()
                      function but it was not possible. Is there any way
                      to get these information? Also, will be possible
                      to get the response plots for this same object?</div>
                    <div
                      style="font-style:normal;font-size:24px;background-color:transparent;font-family:'times
                      new roman','new york',times,serif"><br>
                    </div>
                    <div
                      style="font-style:normal;font-size:24px;background-color:transparent;font-family:'times
                      new roman','new york',times,serif">Cheers,</div>
                    <div> </div>
                    <div>Pablo Riul<br>
                      <br>
                      Departamento de Engenharia e Meio Ambiente<br>
                      Centro de Ciências Aplicadas e Educação<br>
                      Universidade Federal da Paraíba - Campus IV<br>
                      R: Mangueira s/n, Centro CEP: 58.297-000<br>
                      Rio Tinto - Paraíba - Brasil</div>
                    <div><br>
                    </div>
                    <div style="font-size:18pt;font-family:'times new
                      roman','new york',times,serif">
                      <div style="font-size:12pt;font-family:'times new
                        roman','new york',times,serif">
                        <div dir="ltr"> <font face="Arial">
                            <hr size="1"> <b><span
                                style="font-weight:bold">De:</span></b>
                            Damien Georges <a moz-do-not-send="true"
                              href="mailto:damien.georges2@gmail.com"
                              target="_blank"><damien.georges2@gmail.com></a><br>
                            <b><span style="font-weight:bold">Para:</span></b>
                            <a moz-do-not-send="true"
                              href="mailto:biomod-commits@lists.r-forge.r-project.org"
                              target="_blank">biomod-commits@lists.r-forge.r-project.org</a>;
                            Pablo Riul <a moz-do-not-send="true"
                              href="mailto:pabloriul@yahoo.com.br"
                              target="_blank"><pabloriul@yahoo.com.br></a>
                            <br>
                            <b><span style="font-weight:bold">Enviadas:</span></b>
                            Quarta-feira, 12 de Setembro de 2012 10:52<br>
                            <b><span style="font-weight:bold">Assunto:</span></b>
                            Re: [Biomod-commits] Warning in
                            BIOMOD_FormatingData<br>
                          </font> </div>
                        <br>
                        <div>
                          <div>
                            <div>Dear Pablo,<br>
                              <br>
                              Don't worry about it.. It's just a warning
                              indicating that some function of raster
                              package have changed of names.. <br>
                              I took this into account in the last
                              biomod2 version so warnings will disappear
                              when you'll update the package .<br>
                              <br>
                              Best,<br>
                              <br>
                              Damien <br>
                              <br>
                              On 12/09/2012 14:46, Pablo Riul wrote:<br>
                            </div>
                            <blockquote type="cite">
                              <div
                                style="font-size:18pt;font-family:'times
                                new roman','new york',times,serif">
                                <div><span>Dear list,</span></div>
                                <div
                                  style="font-style:normal;font-size:24px;background-color:transparent;font-family:'times
                                  new roman','new york',times,serif"><span><br>
                                  </span></div>
                                <div
                                  style="font-style:normal;font-size:24px;background-color:transparent;font-family:'times
                                  new roman','new york',times,serif"><span>I
                                    updated my packages and I'm now
                                    using raster 2.0-12. When I created
                                    myBiomodData using the
                                    BIOMOD_FormatingData function I
                                    received the warning below:</span></div>
                                <div
                                  style="font-style:normal;font-size:24px;background-color:transparent;font-family:'times
                                  new roman','new york',times,serif"><span><br>
                                  </span></div>
                                <div
                                  style="font-style:normal;font-size:24px;background-color:transparent;font-family:'times
                                  new roman','new york',times,serif"><span>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px
                                      Monaco;color:#1c39f6"><span
                                        style="color:#99228d">> </span>myBiomodData

                                      <-
                                      BIOMOD_FormatingData(resp.var =
                                      myResp,expl.var = myExpl,resp.xy =
                                      myRespXY,<a moz-do-not-send="true"
                                        href="http://resp.name"
                                        target="_blank">resp.name</a> =
                                      myRespName,PA.nb.rep =
                                      1,PA.nb.absences =
                                      1000,PA.strategy = 'random')</div>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px
                                      Monaco;min-height:19.0px"><br>
                                    </div>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px Monaco">-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=


                                      Mhisp Data Formating
                                      -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=</div>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px
                                      Monaco;min-height:19.0px"><br>
                                    </div>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px Monaco">      !
                                      No data has been set aside for
                                      modeling evaluation</div>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px Monaco">   >
                                      Pseudo Absences Selection
                                      checkings...</div>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px Monaco">   >
                                      random pseudo absences selection</div>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px Monaco">   >
                                      Pseudo absences are selected in
                                      explanatory variables</div>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px Monaco">-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=


                                      Done
                                      -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=</div>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px
                                      Monaco;color:#ff1f17">Warning
                                      message:</div>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px
                                      Monaco;color:#ff1f17">In .local(x,
                                      rcl, ...) :</div>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px
                                      Monaco;color:#ff1f17">  raster
                                      function "reclass" has been
                                      deprecated and will be removed
                                      from the package;</div>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px
                                      Monaco;color:#ff1f17">Please use
                                      "reclassify" instead</div>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px
                                      Monaco;color:#ff1f17"><br>
                                    </div>
                                    <div style="margin:0.0px 0.0px 0.0px
                                      0.0px;font:14.2px
                                      Monaco;color:#ff1f17"><br>
                                    </div>
                                  </span></div>
                                <div>Its not a problem now, but maybe in
                                  next updates of package raster the
                                  reclass function may be not available
                                  anymore.</div>
                                <div><br>
                                </div>
                                <div
                                  style="font-style:normal;font-size:24px;background-color:transparent;font-family:'times
                                  new roman','new york',times,serif">Cheers,</div>
                                <div
                                  style="font-style:normal;font-size:24px;background-color:transparent;font-family:'times
                                  new roman','new york',times,serif"> </div>
                                <div>Pablo Riul<br>
                                  <br>
                                  Departamento de Engenharia e Meio
                                  Ambiente<br>
                                  Centro de Ciências Aplicadas e
                                  Educação<br>
                                  Universidade Federal da Paraíba -
                                  Campus IV<br>
                                  R: Mangueira s/n, Centro CEP:
                                  58.297-000<br>
                                  Rio Tinto - Paraíba - Brasil</div>
                              </div>
                              <br>
                              <fieldset></fieldset>
                              <br>
                              <pre>_______________________________________________
Biomod-commits mailing list
<a moz-do-not-send="true" rel="nofollow" href="mailto:Biomod-commits@lists.r-forge.r-project.org" target="_blank">Biomod-commits@lists.r-forge.r-project.org</a>
<a moz-do-not-send="true" rel="nofollow" href="https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/biomod-commits" target="_blank">https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/biomod-commits</a></pre>
                            </blockquote>
                            <br>
                          </div>
                        </div>
                        <br>
                        <br>
                      </div>
                    </div>
                  </div>
                </blockquote>
                <br>
              </div>
            </div>
          </div>
          <br>
          _______________________________________________<br>
          Biomod-commits mailing list<br>
          <a moz-do-not-send="true"
            href="mailto:Biomod-commits@lists.r-forge.r-project.org">Biomod-commits@lists.r-forge.r-project.org</a><br>
          <a moz-do-not-send="true"
href="https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/biomod-commits"
            target="_blank">https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/biomod-commits</a><br>
        </blockquote>
      </div>
      <br>
    </blockquote>
    <br>
  </body>
</html>