Tuning SVM kernel. Trains SVMs with a range of kernels (linear, polynomial degree 2, 3 and 4, radial and sigmoid) using cross validation so the optimal kernel can be chosen (using the resulting plots). If specified (by showplots=F) the plots are saved as jpegs.

selectsvmkernel(data, title, showplots = T, output_prefix = "")

Arguments

data	Dataset: dataframe containing classification column and all other column features. Both the training and test datasets will be taken from this dataset.
title	Title to be used for the resulting boxplot
showplots	T if plots should be shown in standard output, F is plots should be saved as jpg files.
output_prefix	Prefix used for saving plots. If showplots==F then plots are saved here. Otherwise, standard output.

Value

Dataframe containing test and training accuracy, sensitivity and specificity

Examples

data = read.csv(paste(system.file('samples/subsamples', package = "feamiR"),'/sample0.csv',sep=''))
data = rbind(head(data,50),tail(data,50))
data$classification = as.factor(data$classification)
data = data[,2:ncol(data)]
selectsvmkernel(data,'SVM boxplots')
#>     kernel  accuracy sensitivity specificity  type
#> 1   linear 0.8571429   0.0000000   0.7272727  test
#> 2   linear 1.0000000   0.0000000   0.0000000 train
#> 3   radial 0.7619048   0.6000000   0.9090909  test
#> 4   radial 0.8481013   0.7000000   0.0000000 train
#> 5  sigmoid 0.9047619   0.0000000   0.8181818  test
#> 6  sigmoid 0.8734177   0.9500000   0.7948718 train
#> 7   poly 2 0.4761905   0.0000000   0.0000000  test
#> 8   poly 2 0.5063291   0.0000000   0.0000000 train
#> 9   poly 3 0.4761905   0.0000000   0.0000000  test
#> 10  poly 3 0.5063291   0.0000000   0.0000000 train
#> 11  poly 4 0.4761905   0.0000000   0.0000000  test
#> 12  poly 4 0.5063291   0.0000000   0.0000000 train
#> 13  linear 0.9047619   0.0000000   0.7142857  test
#> 14  linear 0.9873418   0.0000000   0.9767442 train
#> 15  radial 0.6666667   0.5000000   0.0000000  test
#> 16  radial 0.7215190   0.3888889   0.0000000 train
#> 17 sigmoid 0.6666667   0.5000000   0.0000000  test
#> 18 sigmoid 0.7215190   0.3888889   0.0000000 train
#> 19  poly 2 0.3333333   0.0000000   0.0000000  test
#> 20  poly 2 0.5443038   0.0000000   0.0000000 train
#> 21  poly 3 0.3333333   0.0000000   0.0000000  test
#> 22  poly 3 0.5443038   0.0000000   0.0000000 train
#> 23  poly 4 0.3333333   0.0000000   0.0000000  test
#> 24  poly 4 0.5443038   0.0000000   0.0000000 train
#> 25  linear 0.8666667   0.8750000   0.8571429  test
#> 26  linear 0.9882353   0.9761905   0.0000000 train
#> 27  radial 0.9333333   0.8750000   0.0000000  test
#> 28  radial 0.9176471   0.8333333   0.0000000 train
#> 29 sigmoid 0.8666667   0.7500000   0.0000000  test
#> 30 sigmoid 0.8352941   0.6666667   0.0000000 train
#> 31  poly 2 0.4666667   0.0000000   0.0000000  test
#> 32  poly 2 0.5058824   0.0000000   0.0000000 train
#> 33  poly 3 0.4666667   0.0000000   0.0000000  test
#> 34  poly 3 0.5058824   0.0000000   0.0000000 train
#> 35  poly 4 0.4666667   0.0000000   0.0000000  test
#> 36  poly 4 0.5058824   0.0000000   0.0000000 train
#> 37  linear 0.7916667   0.9000000   0.7142857  test
#> 38  linear 0.9868421   0.0000000   0.9722222 train
#> 39  radial 0.7916667   0.9000000   0.7142857  test
#> 40  radial 0.9210526   0.9250000   0.9166667 train
#> 41 sigmoid 0.5000000   0.0000000   0.1428571  test
#> 42 sigmoid 0.6842105   0.0000000   0.3333333 train
#> 43  poly 2 0.4166667   0.0000000   0.0000000  test
#> 44  poly 2 0.5263158   0.0000000   0.0000000 train
#> 45  poly 3 0.4166667   0.0000000   0.0000000  test
#> 46  poly 3 0.5263158   0.0000000   0.0000000 train
#> 47  poly 4 0.4166667   0.0000000   0.0000000  test
#> 48  poly 4 0.5263158   0.0000000   0.0000000 train
#> 49  linear 0.7368421   0.8750000   0.6363636  test
#> 50  linear 0.9876543   0.0000000   0.9743590 train
#> 51  radial 0.7894737   0.5000000   0.0000000  test
#> 52  radial 0.9506173   0.9047619   0.0000000 train
#> 53 sigmoid 0.6842105   0.5000000   0.8181818  test
#> 54 sigmoid 0.9012346   0.9047619   0.8974359 train
#> 55  poly 2 0.4210526   0.0000000   0.0000000  test
#> 56  poly 2 0.5185185   0.0000000   0.0000000 train
#> 57  poly 3 0.4210526   0.0000000   0.0000000  test
#> 58  poly 3 0.5185185   0.0000000   0.0000000 train
#> 59  poly 4 0.4210526   0.0000000   0.0000000  test
#> 60  poly 4 0.5185185   0.0000000   0.0000000 train