These problems could be remedied by using
transformation such as exponential, logarithm and others
but sometimes, even after the transformation, problems
with non normal data still occur. Simple transformations
of the data such as by taking logarithm can reduce
skewness but not for complex transformations such as the
class of Box-Cox transformations [18]. However,
problems due to outliers still exist. According to [18], a
simple transformation can alter skewed distributions to
make them more symmetrical, but the approach does
directly eliminate outliers.