Geology Reference
In-Depth Information
Fig. 6.27 Variation of performance in daily rainfall runoff modelling at the Brue catchment when
applying different support vector machines on different kernel functions
Bird Creek region in the USA, where the
-SVM with linear kernel
function have shown better value (minimum) of error (Bray and Han [ 4 ]. The mean
square error produced by the
ε
-SVM and
ν
ε
-SVM was comparable to that of
ν
-SVM in all kernel
functions and the minimum value observed when
ε
-SVM applied with the linear
kernel.
The SVM hypothesis suggests that the performance of SVM depends on the
slack parameter (
ε
) and the cost factor (C). We have performed the analysis varying
the
= 0.00001 and the cost parameters C = 0.1 to
C = 1000. The analysis results are shown in Figs. 6.28 and 6.29 respectively.
Figure 6.28 has shown that the least error increases rapidly for
ε
values between
ε
=1to
ε
ε
greater than 0.1. So
we set the value of
to 0.1 for reliable results and less computation time.
The cost factor of error (C) assigns a penalty for the number of vectors falling
between the two hyperplanes in the hypothesis. It suggests that, if the data is of
good quality the distance between the two hyperplanes is narrowed down. If the
data is noisy it is preferable to have a smaller value of C which will not penalise the
ε
 
Search WWH ::




Custom Search