Some InfoNest stuff to measure entropies in cricket. This depends on InfoNest, which I've included as a submodule in this repository. Make sure you use --recursive when you clone this repository:

$ git clone --recursive https://github.com/eggplantbren/CricketInfo

You'll also need to compile InfoNest before you can compile CricketInfo.

$ cd CricketInfo
$ cd InfoNest/cpp
$ make
$ cd ../../cpp
$ make

(c) 2017 Brendon J. Brewer

This is free software released under the GNU General Public License, version 3. See LICENSE file for details.

These are all for datasets of 100 innings.

For the constant hazard model, the average depth is 8.607 +- 0.044, so the conditional (differential) entropy is 2.393 +- 0.044 nats. The differential entropy of the prior is 4.350 nats. Therefore, the mutual information is 1.957 +- 0.044 nats.

For the exponential hazard model, mu2 was the target parameter. The average depth is 9.275 +- 0.053, so the conditional (differential) entropy is 3.060 +- 0.053 nats. Therefore, the mutual information is 1.290 +- 0.053 nats.

N=10 => Mutual information = 0.910 +- 0.043 nats. N=100 => Mutual information = 1.957 +- 0.044 nats. N=200 => Mutual information = 2.249 +- 0.045 nats.