%0 DATA
%A Petros, Boutselis
%A Trevor, Ringrose
%D 2016
%T COMBAT SIMULATION - TWO SIDES.xls
%U https://cord.cranfield.ac.uk/articles/dataset/COMBAT_SIMULATION_-_TWO_SIDES_xls/3398521
%R 10.17862/cranfield.rd.3398521.v1
%2 https://cord.cranfield.ac.uk/ndownloader/files/5309626
%K DATA FOR THE DEVELOPMENT OF METAMODELS
%K Artificial Neural Networks
%K Statistical models
%K GAMLSS
%K Bayesian Networks
%X The data are the outputs of the simulation of a closed-loop, two-sides (Blue vs Red) combat model saved in an Excel spreadsheet. The 1st tab has the input data while the other two tabs have the outputs for the Blue and for the Red. Each column in the Input tab is a design point randomly selected (250 in total). The first two rows represent the number of tanks and recce of the Blue, while rows 3 and 4 represent the respective Red. The 5th row represents the probability of shock that has been applied the same to both sides. Rows 6, 7 and 8 represent the percentage Unit participation of Tanks, Machine Guns and Anti-tanks respectively. These last three variables also apply equally to both Blue and Red.The model used (SIMBAT) is stochastic and therefore each design point was run 40 times, taking care to use the same random numbers for each point (Common Random Numbers) Therefore each of the 250 columns in the two output tabs has 40 rows.

The data have been used in:

P. Boutselis, Trevor J. Ringrose

GAMLSS and neural networks in combat simulation metamodelling: A case study. Expert Syst. Appl. 40(15): 6087-6093 (2013),

doi:10.1016/j.eswa.2013.05.023to produce two different metamodels: a statistical model (GAMLSS) and a neural network (ANN), while recently the same data have been used to produce a Bayesian Network.