scale the input data for correct learning
[scaled_instance,scaled_parameters] = svmnormalize(instance); [scaled_instance,scaled_parameters] = svmnormalize(instance,[meanV,stdV]); [scaled_instance] = svmnormalize(instance,scaled_parameters); [scaled_instance,scaled_parameters,scaled_label,scaled_label_parameters] = svmnormalize(instance,[meanV,stdV],label,[label_mean, label_std]);
Scale your data. For example, scale each attribute to a mean of 0 and a standard deviation of 1.
[label,instance]=libsvmread("demos/heart_scale"); [scaled_instance,scaled_parameters] = svmnormalize(instance,[0,1]); cc = svmtrain(label,scaled_instance); [predicted_label]=svmtrain(label,svmnormalize(instance,scaled_parameters));