summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authormguetlein <martin.guetlein@gmail.com>2011-08-13 14:20:18 +0200
committermguetlein <martin.guetlein@gmail.com>2011-08-13 14:20:18 +0200
commit1f4ebe46790443225beff01b9e1918f5baec31f8 (patch)
treeda0ce3e558bd27ab6dec20aaf4ca7fd318339671
parentcd07317a2d5130fb9bf8b6ed119aa1e6166810b0 (diff)
add bootstrapping description
-rwxr-xr-xvalidation/validation_application.rb9
1 files changed, 8 insertions, 1 deletions
diff --git a/validation/validation_application.rb b/validation/validation_application.rb
index d2dfef0..0658309 100755
--- a/validation/validation_application.rb
+++ b/validation/validation_application.rb
@@ -400,7 +400,14 @@ get '/bootstrapping' do
"All validations: "+url_for("/",:full)+"\n"+
"Validation reports: "+url_for("/report/validation",:full)
description =
- "A list of all bootstrapping-validations.\n"+
+ "A list of all bootstrapping-validations.\n\n"+
+ "Bootstrapping performs sampling with replacement to create a training dataset and test dataset from the orignial dataset.\n"+
+ "Subsequently, a model is built with the training dataset and validated on the test-dataset.\n\n"+
+ "Quote from R Kohavi - A study of cross-validation and bootstrap for accuracy estimation and model selection,\n"+
+ "International joint Conference on artificial intelligence, 1995:\n"+
+ "'Given a dataset of size n, a bootstrap sample is created by sampling n instances uniformly from the data (with replacement).\n"+
+ " Since the dataset is sampled with replacement, the probability of any given instance not being chosen after n samples is (1 - 1/n)^n = e^-1 = 0.368;\n"+
+ " the expected number of distinct instances from the original dataset appearing in the test set is thus 0.632n.'\n\n"+
"To perform a bootstrapping-validation use the POST method."
post_command = OpenTox::PostCommand.new request.url,"Perform bootstrapping-validation"
post_command.attributes << OpenTox::PostAttribute.new("algorithm_uri")