Hint : Forecasting usually employ 'windowing' on data set,
i.e. given the n previous data points predict the next data point.
Therefore the suggested terminal set is { x1, x2, x3, x4 } where
x1.. x4 is the data of i-1, i-2 ... from the date to be predicted.
That is
predicted value of
day i = f ( i-1, i-2, i-3, i-4 )
As the function should contain some non-linear function, the suggested
function set is { +, -, * , / , sin, cos, exp }. The
data set is given in this format (as text ) ordered from the most recent
to the oldest data (value of bath per us-dollar).
50.5824
52.4113
53.2604
53.3893
54.1132
. . .
Fitness function is simply the root mean square error (RMS) of the predicted
value from the actual value. You must keep some value unseen and
use the rest for fitness cases. The unseen values are used to test the
accuracy of your prediction function.