<div dir="ltr">Senhores, bom dia!<div><br></div><div>Considerando o quarto item colocado pelo Prof. Paulo Justiniano, sobre a real necessidade da validação cruzada, talvez tenham interesse no texto de Davis (1987), intitulado "Uses and abuses of cross-validation in geostatistics".</div>
<div><br></div><div>Obviamente há referências mais atuais, mas essa aí é bastante didática porque, além de tudo, esclarece diferenças entre "cross validation" e "jacknifing".</div><div><br></div><div>Segue um excerto do texto original:<br>
</div><div><br></div><div><div>"Cross-validation  cannot  confirm  that  a  particular  model  is  or  is  not  the  optimum.  It  is  a method  to  better  examine  and  understand  the  phenomenon  under  study  using the  available  data."</div>
<div><br><div><div>Davis, B. M., 1987. Uses and abuses of cross-validation in geostatistics. Math. Geol., v. 17, p. 563–586.<br></div><div><div><div><br></div></div><div><div class="gmail_extra"><br clear="all"><div><div dir="ltr">
Éder Comunello <<a href="mailto:comunello.eder@gmail.com" target="_blank">c</a><a href="mailto:omunello.eder@gmail.com" target="_blank">omunello.eder@gmail.com</a>> <br>Dourados, MS - [22 16.5'S, 54 49'W]<br>
</div></div>
<br><br></div></div></div></div></div></div></div>