search

ag-csw / LDStreamHMMLearn

1 stars 0 forks source link

Percentiles Bayes Error Plot for MM #43

Open alexlafleur opened 7 years ago

alexlafleur commented 7 years ago

taumeta: 4 eta: 16 scale_window : 16 shift: 64 window_size: 1024 num_estimations: 18 len_trajectory: 2177 num_trajectories: 4 num_trajectorieslen_trajectory: 8708 NAIVE window_size num_estimations 19456 BAYES window_size + num_estimations*shift 2176

numruns=8 num_trajectories (simulated) = 128 numsims = 32 num_trajs = 4

Deciles_Bayes_MM.pdf

Issue follows up on #31

greenTara commented 7 years ago

In this plot the prediction doesn't look as good because the error at k=0 is somewhat off, and that throws everything else off. If there is time, let's increase the sample from each model: numsims = 64.

If we had all the time in the world to finish this, we would fit the constant by taking the ratio of the mean error[k] divided by the formula (without the constant) and then average to get the best-fit value for the constant. But other things are higher priority now.

alexlafleur commented 7 years ago

Deciles_Bayes_MM.pdf evaluate_deciles_mm.txt