False Discovery Rates and C - Statistics

[iamciera]

NEEDS to be done

• [x] We need a better way to assess accuracy. It is not all about getting the highest validation accuracy. We need to get the false discovery rates and c-statistics. We need to build this into our scripts, create confusion matrices, and create documentation.
• [x] Present to group the code and what the statistics mean

[iamciera]

Can you run on: https://github.com/DiscoveryDNA/team_neural_network/blob/master/code/2018-07-29_pointing_out_a_fault_in_previous_experiments.ipynb

[zhanyuanucb]

A script for calculating the FDR and C-statistics of a model: https://github.com/DiscoveryDNA/team_neural_network/blob/master/code/utility/FDR_and_C_stats.ipynb

Apply the FDR-C-statistics script on a model: https://github.com/DiscoveryDNA/team_neural_network/blob/master/code/729model_FDR_and_C_stats.ipynb