Consider a data set D with 10 data points {-1,6,0,2,-1,7,7,8,4,-2}. We want to find a model for M from a restricted sample space Ω={0,2,4}. Assume the data has Laplace noise defined, so from a model M a data point's probability distribution
is described as f(x)=1/4exp(-|M-x 1/2). Also assume we have a prior assumption on the models so that Pr(M=0)=0.25,Pr(M=2)=0.35, and Pr(M=4)=0.4. Assuming all data points in D are independent, which model is most likely?
