Bronchopulmonary Dysplasia Prediction Using Support Vector Machine and Logit Regression

Bronchopulmonary Dysplasia Prediction Using Support Vector Machine and LIBSVM

“…To compare, in each presented model mean value of ACC, T P R and SP C were obtained with different Jacknife parameters, using both methods: LR and sigmoid SVM with LIBSVM. Where applicable we added RBF Matlab SVM implementation results (as M. SVM) from [2].…”

“…The best choice for such a prediction is to use the LIBSVM instead of Matlab's implementation, which gives less control on computation process. Most likely that was the reason why the bigger parameter set we used the worse results we got using Matlab [2]. Although we did not test all possible 2 14 combinations of parameters (only 3375 random models), nonetheless looking on Table III it can be concluded that standard deviation of accuracy for SVM is much higher than for logit regression.…”

“…To compare with LR algorithm we used data from previous article [2], where we reviewed all of the 2 14 possible combinations of models with exactly same Jacknife and crossvalidation procedure.…”

“…Many of authors mention use of support vector machine(SVM) [27] in future works, however it is difficult to find them. In our previous work [2] we compared SVM with LR classifiers. Unfortunately, due to internal Matlab SVM library usage results were highly unsatisfactory.…”

“…SVM (Support Vector Machine) algorithm implemented in LIBSVM[1] was used as classifier. Results are compared to others gathered in previous work [2] where LR (Logit Regression) and Matlab environment SVM implementation were used. Fourteen different risk factor parameters were considered and due to the high computational complexity only 3375 random combinations were analysed.…”