many thanks victor ! On Wednesday, October 26, 2016 at 3:48:18 PM UTC+2, Poughon Victor wrote: > > No worries! > > You could also use 100% of labels for training. Of course this means that > you cannot do the testing step. > The testing is there to evaluate the performance and quality of learned > model. It is mostly used when developing new method, or tuning some > parameters. It is not required if you just want the classification results. > But without it, you will be unable to tell if it's a good one or not! > > Some people even split the reference data into 3 subsets: training, > validation and testing, for various reasons. > See this interesting page: http://stats.stackexchange.com/a/19051 > > There are other way to evaluate a classification model (leave-one-out, > cross validation, etc.), but a 70/30 split into training and testing is a > simple and common one. > > Victor Poughon > ------------------------------ > *De :* [email protected] <javascript:> [[email protected] > <javascript:>] de la part de michel M [[email protected] <javascript:>] > *Envoyé :* mercredi 26 octobre 2016 14:13 > *À :* otb-users > *Objet :* [otb-users] Re: exercice corrigé (training.shp et testing.shp) > > Thanks victor but for me it's not again clear ... > For exemple I have a shape file with data source corine land cover , i > have a crop code , a label and a geometry > I put 70% of records in a testing file > and 30% of others records in a training file > and after i don't understand the mécanism .... > why not just all records in a training shape file > many thanks > > > On Monday, October 24, 2016 at 5:54:38 PM UTC+2, michel M wrote: >> >> Bonjour, >> Suite au dernier corrigé de Montpellier (Theia) , >> a partir de la page 14 classification supervisée pour les séries >> multitemporelles >> Je ne comprends pas la difference entre training et testing ) >> en effet je suis en train de faire un petit test ( téléchargement de 9 >> images landsat sur THEIA de l'année 2014 , déja orthorectifié ... et >> corriger au niveau atmosphérique ) pour visualiser les cultures sur une >> zone... >> >> Dans les corrigés >> ils ont 2 fichiers shapes training.shp et testing.shp >> -- training contient la donnée d'apprentissage >> -- testing contient la donnée de validation >> ( je comprends bien que training c'est les zones d'entraînement choisi >> par l'utilisateur ) >> et testing c'est quoi ??) >> >> - si par exemple je vais créer un fichier shape provenant des cultures >> de RPG2014 (recensement parcellaire graphique ) , je vais le mettre en >> training est ce OK ? >> >> mais testing ( je ne vois pas ) >> >> Pouvez vous m'éclairer ? >> >> merci >> Michel >> >> -- > -- > Check the OTB FAQ at > http://www.orfeo-toolbox.org/FAQ.html > > You received this message because you are subscribed to the Google > Groups "otb-users" group. > To post to this group, send email to [email protected] > <javascript:> > To unsubscribe from this group, send email to > [email protected] <javascript:> > For more options, visit this group at > http://groups.google.com/group/otb-users?hl=en > --- > You received this message because you are subscribed to the Google Groups > "otb-users" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected] <javascript:>. > For more options, visit https://groups.google.com/d/optout. >
-- -- Check the OTB FAQ at http://www.orfeo-toolbox.org/FAQ.html You received this message because you are subscribed to the Google Groups "otb-users" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/otb-users?hl=en --- You received this message because you are subscribed to the Google Groups "otb-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
