Friday, February 24, 2017

BTO Error Scatter Plot

From "Bayesian Methods..." page 27. The computed standard deviation value of these R1200 samples is 29 sec. So 4-sigma would be ~120 usec.  At least 11 samples below exceed 4-sigma.






















A scatterplot of 30,000 Gaussian distributed samples was created and is shown below. The probability of a 4-sigma sample is a little over one in 15,000. It is not known how many samples are in the DSTG BTO scatter plot, but it is surely not more than a thousand or so. In the 30,000 Gaussian sample below only one sample exceeded 4-sigma and the 3-sigma values are in the range of 3 in 1,000 samples. The probability of a 3-sigma value is about one in 370.





























So basically the BTO is not strictly Gaussian distributed, and large BTO values are over represented in the sampled data.