Skip to main content
. 2019 Nov 29;19(23):5255. doi: 10.3390/s19235255
Algorithm 1. Pseudocode for modeling process using clustered dataset.
EXPERIMENTS = 10;
TOTAL_POTS = 960;
POTS_BY_SECTION = 30;
TOTAL_OUTPUTS = 3;
for i_exp = 1 to EXPERIMENTS do
 for i_out = 1 to TOTAL_OUTPUTS do
  for i_pot = 1 to 30 to TOTAL_POTS do
   a) Get data from a section:
    (index_pot >= i_pot and index_pot <= (i_pot + POTS_BY_SECTION − 1).
   b) Create input and output (i_out) data matrices.
   c) Split data between training and validation datasets.
   d) Define parameters of the ANN model.
   e) Create ANN model.
   f) Train ANN model.
   for i_test = i_pot to (i_pot + POTS_BY_SECTION − 1) do
    g) Get data by index_pot = i_test.
    h) Create input and output (i_out) data matrices.
    i) Simulate ANN model using data by (step h)).
    j) Calculate and store MSE and R values.
    k) Check if MSE and R values are better than previous model. If true, store model.
   end_for
  end_for
 end_for
end_for
print/plot MSEtest values by each experiments and output variable
print/plot Rtest values by each experiments and output variable
l) Calculate MSEtest and Rtest average:
print MSEglobal by each output variable
print Rglobal by each output variable