SFA_STEP Update a step of the SFA algorithm.
SFA_STEP(HLD, DATA[, STEP]) updates the current step of the SFA
algorithm. HDL is the handle of an SFA object, DATA contains a chunk of
input data (each column a different variable).
STEP specifies the current SFA step. If not specified, the current
step is used. The steps must be given in the right sequence:
for SFA1 objects: 'preprocessing', 'sfa'
for SFA2 objects: 'preprocessing', 'expansion', 'sfa'
Each time a new step is invoked, the previous one is closed, which
might take some time.
Example: suppose you have divided your training data into two chunks,
DATA1 and DATA2. Let the number of input dimensions be N. To apply
SFA on them write:
hdl = sfa2_create(N,xp_dim(N))
sfa_step(hdl, DATA1, 'preprocessing')
sfa_step(hdl, DATA2)
sfa_step(hdl, DATA1, 'expansion')
sfa_step(hdl, DATA2)
sfa_step(hdl, [], 'sfa')
output1 = sfa_execute(hdl, DATA1)
output2 = sfa_execute(hdl, DATA2)
See also SFA1_CREATE, SFA2_CREATE, SFA_EXECUTE